linux/drivers/base/core.c

4994 lines
136 KiB
C
Raw Normal View History

// SPDX-License-Identifier: GPL-2.0
/*
* drivers/base/core.c - core driver model code (device registration, etc)
*
* Copyright (c) 2002-3 Patrick Mochel
* Copyright (c) 2002-3 Open Source Development Labs
* Copyright (c) 2006 Greg Kroah-Hartman <gregkh@suse.de>
* Copyright (c) 2006 Novell, Inc.
*/
#include <linux/acpi.h>
#include <linux/cpufreq.h>
#include <linux/device.h>
#include <linux/err.h>
2015-04-03 21:23:37 +00:00
#include <linux/fwnode.h>
#include <linux/init.h>
#include <linux/module.h>
#include <linux/slab.h>
#include <linux/string.h>
#include <linux/kdev_t.h>
Driver core: add notification of bus events I finally did as you suggested and added the notifier to the struct bus_type itself. There are still problems to be expected is something attaches to a bus type where the code can hook in different struct device sub-classes (which is imho a big bogosity but I won't even try to argue that case now) but it will solve nicely a number of issues I've had so far. That also means that clients interested in registering for such notifications have to do it before devices are added and after bus types are registered. Fortunately, most bus types that matter for the various usage scenarios I have in mind are registerd at postcore_initcall time, which means I have a really nice spot at arch_initcall time to add my notifiers. There are 4 notifications provided. Device being added (before hooked to the bus) and removed (failure of previous case or after being unhooked from the bus), along with driver being bound to a device and about to be unbound. The usage I have for these are: - The 2 first ones are used to maintain a struct device_ext that is hooked to struct device.firmware_data. This structure contains for now a pointer to the Open Firmware node related to the device (if any), the NUMA node ID (for quick access to it) and the DMA operations pointers & iommu table instance for DMA to/from this device. For bus types I own (like IBM VIO or EBUS), I just maintain that structure directly from the bus code when creating the devices. But for bus types managed by generic code like PCI or platform (actually, of_platform which is a variation of platform linked to Open Firmware device-tree), I need this notifier. - The other two ones have a completely different usage scenario. I have cases where multiple devices and their drivers depend on each other. For example, the IBM EMAC network driver needs to attach to a MAL DMA engine which is a separate device, and a PHY interface which is also a separate device. They are all of_platform_device's (well, about to be with my upcoming patches) but there is no say in what precise order the core will "probe" them and instanciate the various modules. The solution I found for that is to have the drivers for emac to use multithread_probe, and wait for a driver to be bound to the target MAL and PHY control devices (the device-tree contains reference to the MAL and PHY interface nodes, which I can then match to of_platform_devices). Right now, I've been polling, but with that notifier, I can more cleanly wait (with a timeout of course). Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2006-10-25 03:44:59 +00:00
#include <linux/notifier.h>
#include <linux/of.h>
#include <linux/of_device.h>
#include <linux/blkdev.h>
#include <linux/mutex.h>
#include <linux/pm_runtime.h>
#include <linux/netdevice.h>
#include <linux/sched/signal.h>
#include <linux/sched/mm.h>
#include <linux/swiotlb.h>
#include <linux/sysfs.h>
#include <linux/dma-map-ops.h> /* for dma_default_coherent */
#include "base.h"
#include "physical_location.h"
#include "power/power.h"
#ifdef CONFIG_SYSFS_DEPRECATED
#ifdef CONFIG_SYSFS_DEPRECATED_V2
long sysfs_deprecated = 1;
#else
long sysfs_deprecated = 0;
#endif
static int __init sysfs_deprecated_setup(char *arg)
{
return kstrtol(arg, 10, &sysfs_deprecated);
}
early_param("sysfs.deprecated", sysfs_deprecated_setup);
#endif
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
/* Device links support. */
driver core: Add sync_state driver/bus callback This sync_state driver/bus callback is called once all the consumers of a supplier have probed successfully. This allows the supplier device's driver/bus to sync the supplier device's state to the software state with the guarantee that all the consumers are actively managing the resources provided by the supplier device. To maintain backwards compatibility and ease transition from existing frameworks and resource cleanup schemes, late_initcall_sync is the earliest when the sync_state callback might be called. There is no upper bound on the time by which the sync_state callback has to be called. This is because if a consumer device never probes, the supplier has to maintain its resources in the state left by the bootloader. For example, if the bootloader leaves the display backlight at a fixed voltage and the backlight driver is never probed, you don't want the backlight to ever be turned off after boot up. Also, when multiple devices are added after kernel init, some suppliers could be added before their consumer devices get added. In these instances, the supplier devices could get their sync_state callback called right after they probe because the consumers devices haven't had a chance to create device links to the suppliers. To handle this correctly, this change also provides APIs to pause/resume sync state callbacks so that when multiple devices are added, their sync_state callback evaluation can be postponed to happen after all of them are added. kbuild test robot reported missing documentation for device.state_synced Reported-by: kbuild test robot <lkp@intel.com> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20190904211126.47518-5-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-09-04 21:11:23 +00:00
static LIST_HEAD(deferred_sync);
static unsigned int defer_sync_state_count = 1;
static DEFINE_MUTEX(fwnode_link_lock);
static bool fw_devlink_is_permissive(void);
driver core: Improve fw_devlink & deferred_probe_timeout interaction deferred_probe_timeout kernel commandline parameter allows probing of consumer devices if the supplier devices don't have any drivers. fw_devlink=on will indefintely block probe() calls on a device if all its suppliers haven't probed successfully. This completely skips calls to driver_deferred_probe_check_state() since that's only called when a .probe() function calls framework APIs. So fw_devlink=on breaks deferred_probe_timeout. deferred_probe_timeout in its current state also ignores a lot of information that's now available to the kernel. It assumes all suppliers that haven't probed when the timer expires (or when initcalls are done on a static kernel) will never probe and fails any calls to acquire resources from these unprobed suppliers. However, this assumption by deferred_probe_timeout isn't true under many conditions. For example: - If the consumer happens to be before the supplier in the deferred probe list. - If the supplier itself is waiting on its supplier to probe. This patch fixes both these issues by relaxing device links between devices only if the supplier doesn't have any driver that could match with (NOT bound to) the supplier device. This way, we only fail attempts to acquire resources from suppliers that truly don't have any driver vs suppliers that just happen to not have probed yet. Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20210402040342.2944858-3-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2021-04-02 04:03:41 +00:00
static bool fw_devlink_drv_reg_done;
driver core: Add wait_for_init_devices_probe helper function Some devices might need to be probed and bound successfully before the kernel boot sequence can finish and move on to init/userspace. For example, a network interface might need to be bound to be able to mount a NFS rootfs. With fw_devlink=on by default, some of these devices might be blocked from probing because they are waiting on a optional supplier that doesn't have a driver. While fw_devlink will eventually identify such devices and unblock the probing automatically, it might be too late by the time it unblocks the probing of devices. For example, the IP4 autoconfig might timeout before fw_devlink unblocks probing of the network interface. This function is available to temporarily try and probe all devices that have a driver even if some of their suppliers haven't been added or don't have drivers. The drivers can then decide which of the suppliers are optional vs mandatory and probe the device if possible. By the time this function returns, all such "best effort" probes are guaranteed to be completed. If a device successfully probes in this mode, we delete all fw_devlink discovered dependencies of that device where the supplier hasn't yet probed successfully because they have to be optional dependencies. This also means that some devices that aren't needed for init and could have waited for their optional supplier to probe (when the supplier's module is loaded later on) would end up probing prematurely with limited functionality. So call this function only when boot would fail without it. Tested-by: Geert Uytterhoeven <geert+renesas@glider.be> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20220601070707.3946847-5-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-06-01 07:07:00 +00:00
static bool fw_devlink_best_effort;
/**
* fwnode_link_add - Create a link between two fwnode_handles.
* @con: Consumer end of the link.
* @sup: Supplier end of the link.
*
* Create a fwnode link between fwnode handles @con and @sup. The fwnode link
* represents the detail that the firmware lists @sup fwnode as supplying a
* resource to @con.
*
* The driver core will use the fwnode link to create a device link between the
* two device objects corresponding to @con and @sup when they are created. The
* driver core will automatically delete the fwnode link between @con and @sup
* after doing that.
*
* Attempts to create duplicate links between the same pair of fwnode handles
* are ignored and there is no reference counting.
*/
int fwnode_link_add(struct fwnode_handle *con, struct fwnode_handle *sup)
{
struct fwnode_link *link;
int ret = 0;
mutex_lock(&fwnode_link_lock);
list_for_each_entry(link, &sup->consumers, s_hook)
if (link->consumer == con)
goto out;
link = kzalloc(sizeof(*link), GFP_KERNEL);
if (!link) {
ret = -ENOMEM;
goto out;
}
link->supplier = sup;
INIT_LIST_HEAD(&link->s_hook);
link->consumer = con;
INIT_LIST_HEAD(&link->c_hook);
list_add(&link->s_hook, &sup->consumers);
list_add(&link->c_hook, &con->suppliers);
pr_debug("%pfwP Linked as a fwnode consumer to %pfwP\n",
con, sup);
out:
mutex_unlock(&fwnode_link_lock);
return ret;
}
/**
* __fwnode_link_del - Delete a link between two fwnode_handles.
* @link: the fwnode_link to be deleted
*
* The fwnode_link_lock needs to be held when this function is called.
*/
static void __fwnode_link_del(struct fwnode_link *link)
{
pr_debug("%pfwP Dropping the fwnode link to %pfwP\n",
link->consumer, link->supplier);
list_del(&link->s_hook);
list_del(&link->c_hook);
kfree(link);
}
/**
* fwnode_links_purge_suppliers - Delete all supplier links of fwnode_handle.
* @fwnode: fwnode whose supplier links need to be deleted
*
* Deletes all supplier links connecting directly to @fwnode.
*/
static void fwnode_links_purge_suppliers(struct fwnode_handle *fwnode)
{
struct fwnode_link *link, *tmp;
mutex_lock(&fwnode_link_lock);
list_for_each_entry_safe(link, tmp, &fwnode->suppliers, c_hook)
__fwnode_link_del(link);
mutex_unlock(&fwnode_link_lock);
}
/**
* fwnode_links_purge_consumers - Delete all consumer links of fwnode_handle.
* @fwnode: fwnode whose consumer links need to be deleted
*
* Deletes all consumer links connecting directly to @fwnode.
*/
static void fwnode_links_purge_consumers(struct fwnode_handle *fwnode)
{
struct fwnode_link *link, *tmp;
mutex_lock(&fwnode_link_lock);
list_for_each_entry_safe(link, tmp, &fwnode->consumers, s_hook)
__fwnode_link_del(link);
mutex_unlock(&fwnode_link_lock);
}
/**
* fwnode_links_purge - Delete all links connected to a fwnode_handle.
* @fwnode: fwnode whose links needs to be deleted
*
* Deletes all links connecting directly to a fwnode.
*/
void fwnode_links_purge(struct fwnode_handle *fwnode)
{
fwnode_links_purge_suppliers(fwnode);
fwnode_links_purge_consumers(fwnode);
}
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
void fw_devlink_purge_absent_suppliers(struct fwnode_handle *fwnode)
driver core: fw_devlink: Detect supplier devices that will never be added During the initial parsing of firmware by fw_devlink, fw_devlink might infer that some supplier firmware nodes would get populated as devices. But the inference is not always correct. This patch tries to logically detect and fix such mistakes as boot progresses or more devices probe. fw_devlink makes a fundamental assumption that once a device binds to a driver, it will populate (i.e: add as struct devices) all the child firmware nodes that could be populated as devices (if they aren't populated already). So, whenever a device probes, we check all its child firmware nodes. If a child firmware node has a corresponding device populated, we don't modify the child node or its descendants. However, if a child firmware node has not been populated as a device, we delete all the fwnode links where the child node or its descendants are suppliers. This ensures that no other device is blocked on a firmware node that will never be populated as a device. We also mark such fwnodes as NOT_DEVICE, so that no new fwnode links are created with these nodes as suppliers. Fixes: e590474768f1 ("driver core: Set fw_devlink=on by default") Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Acked-by: Rafael J. Wysocki <rafael@kernel.org> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20210205222644.2357303-2-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2021-02-05 22:26:37 +00:00
{
struct fwnode_handle *child;
/* Don't purge consumer links of an added child */
if (fwnode->dev)
return;
fwnode->flags |= FWNODE_FLAG_NOT_DEVICE;
fwnode_links_purge_consumers(fwnode);
fwnode_for_each_available_child_node(fwnode, child)
fw_devlink_purge_absent_suppliers(child);
}
EXPORT_SYMBOL_GPL(fw_devlink_purge_absent_suppliers);
driver core: fw_devlink: Detect supplier devices that will never be added During the initial parsing of firmware by fw_devlink, fw_devlink might infer that some supplier firmware nodes would get populated as devices. But the inference is not always correct. This patch tries to logically detect and fix such mistakes as boot progresses or more devices probe. fw_devlink makes a fundamental assumption that once a device binds to a driver, it will populate (i.e: add as struct devices) all the child firmware nodes that could be populated as devices (if they aren't populated already). So, whenever a device probes, we check all its child firmware nodes. If a child firmware node has a corresponding device populated, we don't modify the child node or its descendants. However, if a child firmware node has not been populated as a device, we delete all the fwnode links where the child node or its descendants are suppliers. This ensures that no other device is blocked on a firmware node that will never be populated as a device. We also mark such fwnodes as NOT_DEVICE, so that no new fwnode links are created with these nodes as suppliers. Fixes: e590474768f1 ("driver core: Set fw_devlink=on by default") Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Acked-by: Rafael J. Wysocki <rafael@kernel.org> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20210205222644.2357303-2-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2021-02-05 22:26:37 +00:00
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
#ifdef CONFIG_SRCU
static DEFINE_MUTEX(device_links_lock);
DEFINE_STATIC_SRCU(device_links_srcu);
static inline void device_links_write_lock(void)
{
mutex_lock(&device_links_lock);
}
static inline void device_links_write_unlock(void)
{
mutex_unlock(&device_links_lock);
}
int device_links_read_lock(void) __acquires(&device_links_srcu)
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
{
return srcu_read_lock(&device_links_srcu);
}
void device_links_read_unlock(int idx) __releases(&device_links_srcu)
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
{
srcu_read_unlock(&device_links_srcu, idx);
}
int device_links_read_lock_held(void)
{
return srcu_read_lock_held(&device_links_srcu);
}
static void device_link_synchronize_removal(void)
{
synchronize_srcu(&device_links_srcu);
}
static void device_link_remove_from_lists(struct device_link *link)
{
list_del_rcu(&link->s_node);
list_del_rcu(&link->c_node);
}
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
#else /* !CONFIG_SRCU */
static DECLARE_RWSEM(device_links_lock);
static inline void device_links_write_lock(void)
{
down_write(&device_links_lock);
}
static inline void device_links_write_unlock(void)
{
up_write(&device_links_lock);
}
int device_links_read_lock(void)
{
down_read(&device_links_lock);
return 0;
}
void device_links_read_unlock(int not_used)
{
up_read(&device_links_lock);
}
#ifdef CONFIG_DEBUG_LOCK_ALLOC
int device_links_read_lock_held(void)
{
return lockdep_is_held(&device_links_lock);
}
#endif
static inline void device_link_synchronize_removal(void)
{
}
static void device_link_remove_from_lists(struct device_link *link)
{
list_del(&link->s_node);
list_del(&link->c_node);
}
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
#endif /* !CONFIG_SRCU */
static bool device_is_ancestor(struct device *dev, struct device *target)
{
while (target->parent) {
target = target->parent;
if (dev == target)
return true;
}
return false;
}
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
/**
* device_is_dependent - Check if one device depends on another one
* @dev: Device to check dependencies for.
* @target: Device to check against.
*
* Check if @target depends on @dev or any device dependent on it (its child or
* its consumer etc). Return 1 if that is the case or 0 otherwise.
*/
int device_is_dependent(struct device *dev, void *target)
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
{
struct device_link *link;
int ret;
/*
* The "ancestors" check is needed to catch the case when the target
* device has not been completely initialized yet and it is still
* missing from the list of children of its parent device.
*/
if (dev == target || device_is_ancestor(dev, target))
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
return 1;
ret = device_for_each_child(dev, target, device_is_dependent);
if (ret)
return ret;
list_for_each_entry(link, &dev->links.consumers, s_node) {
if ((link->flags & ~DL_FLAG_INFERRED) ==
(DL_FLAG_SYNC_STATE_ONLY | DL_FLAG_MANAGED))
continue;
if (link->consumer == target)
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
return 1;
ret = device_is_dependent(link->consumer, target);
if (ret)
break;
}
return ret;
}
driver core: Remove device link creation limitation If device_link_add() is called for a consumer/supplier pair with an existing device link between them and the existing link's type is not in agreement with the flags passed to that function by its caller, NULL will be returned. That is seriously inconvenient, because it forces the callers of device_link_add() to worry about what others may or may not do even if that is not relevant to them for any other reasons. It turns out, however, that this limitation can be made go away relatively easily. The underlying observation is that if DL_FLAG_STATELESS has been passed to device_link_add() in flags for the given consumer/supplier pair at least once, calling either device_link_del() or device_link_remove() to release the link returned by it should work, but there are no other requirements associated with that flag. In turn, if at least one of the callers of device_link_add() for the given consumer/supplier pair has not passed DL_FLAG_STATELESS to it in flags, the driver core should track the status of the link and act on it as appropriate (ie. the link should be treated as "managed"). This means that DL_FLAG_STATELESS needs to be set for managed device links and it should be valid to call device_link_del() or device_link_remove() to drop references to them in certain sutiations. To allow that to happen, introduce a new (internal) device link flag called DL_FLAG_MANAGED and make device_link_add() set it automatically whenever DL_FLAG_STATELESS is not passed to it. Also make it take additional references to existing device links that were previously stateless (that is, with DL_FLAG_STATELESS set and DL_FLAG_MANAGED unset) and will need to be managed going forward and initialize their status (which has been DL_STATE_NONE so far). Accordingly, when a managed device link is dropped automatically by the driver core, make it clear DL_FLAG_MANAGED, reset the link's status back to DL_STATE_NONE and drop the reference to it associated with DL_FLAG_MANAGED instead of just deleting it right away (to allow it to stay around in case it still needs to be released explicitly by someone). With that, since setting DL_FLAG_STATELESS doesn't mean that the device link in question is not managed any more, replace all of the status-tracking checks against DL_FLAG_STATELESS with analogous checks against DL_FLAG_MANAGED and update the documentation to reflect these changes. While at it, make device_link_add() reject flags that it does not recognize, including DL_FLAG_MANAGED. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Reviewed-by: Saravana Kannan <saravanak@google.com> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Review-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/2305283.AStDPdUUnE@kreacher Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-07-16 15:21:06 +00:00
static void device_link_init_status(struct device_link *link,
struct device *consumer,
struct device *supplier)
{
switch (supplier->links.status) {
case DL_DEV_PROBING:
switch (consumer->links.status) {
case DL_DEV_PROBING:
/*
* A consumer driver can create a link to a supplier
* that has not completed its probing yet as long as it
* knows that the supplier is already functional (for
* example, it has just acquired some resources from the
* supplier).
*/
link->status = DL_STATE_CONSUMER_PROBE;
break;
default:
link->status = DL_STATE_DORMANT;
break;
}
break;
case DL_DEV_DRIVER_BOUND:
switch (consumer->links.status) {
case DL_DEV_PROBING:
link->status = DL_STATE_CONSUMER_PROBE;
break;
case DL_DEV_DRIVER_BOUND:
link->status = DL_STATE_ACTIVE;
break;
default:
link->status = DL_STATE_AVAILABLE;
break;
}
break;
case DL_DEV_UNBINDING:
link->status = DL_STATE_SUPPLIER_UNBIND;
break;
default:
link->status = DL_STATE_DORMANT;
break;
}
}
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
static int device_reorder_to_tail(struct device *dev, void *not_used)
{
struct device_link *link;
/*
* Devices that have not been registered yet will be put to the ends
* of the lists during the registration, so skip them here.
*/
if (device_is_registered(dev))
devices_kset_move_last(dev);
if (device_pm_initialized(dev))
device_pm_move_last(dev);
device_for_each_child(dev, NULL, device_reorder_to_tail);
list_for_each_entry(link, &dev->links.consumers, s_node) {
if ((link->flags & ~DL_FLAG_INFERRED) ==
(DL_FLAG_SYNC_STATE_ONLY | DL_FLAG_MANAGED))
continue;
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
device_reorder_to_tail(link->consumer, NULL);
}
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
return 0;
}
/**
* device_pm_move_to_tail - Move set of devices to the end of device lists
* @dev: Device to move
*
* This is a device_reorder_to_tail() wrapper taking the requisite locks.
*
* It moves the @dev along with all of its children and all of its consumers
* to the ends of the device_kset and dpm_list, recursively.
*/
void device_pm_move_to_tail(struct device *dev)
{
int idx;
idx = device_links_read_lock();
device_pm_lock();
device_reorder_to_tail(dev, NULL);
device_pm_unlock();
device_links_read_unlock(idx);
}
#define to_devlink(dev) container_of((dev), struct device_link, link_dev)
static ssize_t status_show(struct device *dev,
struct device_attribute *attr, char *buf)
{
const char *output;
switch (to_devlink(dev)->status) {
case DL_STATE_NONE:
output = "not tracked";
break;
case DL_STATE_DORMANT:
output = "dormant";
break;
case DL_STATE_AVAILABLE:
output = "available";
break;
case DL_STATE_CONSUMER_PROBE:
output = "consumer probing";
break;
case DL_STATE_ACTIVE:
output = "active";
break;
case DL_STATE_SUPPLIER_UNBIND:
output = "supplier unbinding";
break;
default:
output = "unknown";
break;
}
return sysfs_emit(buf, "%s\n", output);
}
static DEVICE_ATTR_RO(status);
static ssize_t auto_remove_on_show(struct device *dev,
struct device_attribute *attr, char *buf)
{
struct device_link *link = to_devlink(dev);
const char *output;
if (link->flags & DL_FLAG_AUTOREMOVE_SUPPLIER)
output = "supplier unbind";
else if (link->flags & DL_FLAG_AUTOREMOVE_CONSUMER)
output = "consumer unbind";
else
output = "never";
return sysfs_emit(buf, "%s\n", output);
}
static DEVICE_ATTR_RO(auto_remove_on);
static ssize_t runtime_pm_show(struct device *dev,
struct device_attribute *attr, char *buf)
{
struct device_link *link = to_devlink(dev);
drivers core: Use sysfs_emit and sysfs_emit_at for show(device *...) functions Convert the various sprintf fmaily calls in sysfs device show functions to sysfs_emit and sysfs_emit_at for PAGE_SIZE buffer safety. Done with: $ spatch -sp-file sysfs_emit_dev.cocci --in-place --max-width=80 . And cocci script: $ cat sysfs_emit_dev.cocci @@ identifier d_show; identifier dev, attr, buf; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... return - sprintf(buf, + sysfs_emit(buf, ...); ...> } @@ identifier d_show; identifier dev, attr, buf; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... return - snprintf(buf, PAGE_SIZE, + sysfs_emit(buf, ...); ...> } @@ identifier d_show; identifier dev, attr, buf; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... return - scnprintf(buf, PAGE_SIZE, + sysfs_emit(buf, ...); ...> } @@ identifier d_show; identifier dev, attr, buf; expression chr; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... return - strcpy(buf, chr); + sysfs_emit(buf, chr); ...> } @@ identifier d_show; identifier dev, attr, buf; identifier len; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... len = - sprintf(buf, + sysfs_emit(buf, ...); ...> return len; } @@ identifier d_show; identifier dev, attr, buf; identifier len; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... len = - snprintf(buf, PAGE_SIZE, + sysfs_emit(buf, ...); ...> return len; } @@ identifier d_show; identifier dev, attr, buf; identifier len; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... len = - scnprintf(buf, PAGE_SIZE, + sysfs_emit(buf, ...); ...> return len; } @@ identifier d_show; identifier dev, attr, buf; identifier len; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... - len += scnprintf(buf + len, PAGE_SIZE - len, + len += sysfs_emit_at(buf, len, ...); ...> return len; } @@ identifier d_show; identifier dev, attr, buf; expression chr; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { ... - strcpy(buf, chr); - return strlen(buf); + return sysfs_emit(buf, chr); } Signed-off-by: Joe Perches <joe@perches.com> Link: https://lore.kernel.org/r/3d033c33056d88bbe34d4ddb62afd05ee166ab9a.1600285923.git.joe@perches.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-09-16 20:40:39 +00:00
return sysfs_emit(buf, "%d\n", !!(link->flags & DL_FLAG_PM_RUNTIME));
}
static DEVICE_ATTR_RO(runtime_pm);
static ssize_t sync_state_only_show(struct device *dev,
struct device_attribute *attr, char *buf)
{
struct device_link *link = to_devlink(dev);
drivers core: Use sysfs_emit and sysfs_emit_at for show(device *...) functions Convert the various sprintf fmaily calls in sysfs device show functions to sysfs_emit and sysfs_emit_at for PAGE_SIZE buffer safety. Done with: $ spatch -sp-file sysfs_emit_dev.cocci --in-place --max-width=80 . And cocci script: $ cat sysfs_emit_dev.cocci @@ identifier d_show; identifier dev, attr, buf; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... return - sprintf(buf, + sysfs_emit(buf, ...); ...> } @@ identifier d_show; identifier dev, attr, buf; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... return - snprintf(buf, PAGE_SIZE, + sysfs_emit(buf, ...); ...> } @@ identifier d_show; identifier dev, attr, buf; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... return - scnprintf(buf, PAGE_SIZE, + sysfs_emit(buf, ...); ...> } @@ identifier d_show; identifier dev, attr, buf; expression chr; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... return - strcpy(buf, chr); + sysfs_emit(buf, chr); ...> } @@ identifier d_show; identifier dev, attr, buf; identifier len; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... len = - sprintf(buf, + sysfs_emit(buf, ...); ...> return len; } @@ identifier d_show; identifier dev, attr, buf; identifier len; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... len = - snprintf(buf, PAGE_SIZE, + sysfs_emit(buf, ...); ...> return len; } @@ identifier d_show; identifier dev, attr, buf; identifier len; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... len = - scnprintf(buf, PAGE_SIZE, + sysfs_emit(buf, ...); ...> return len; } @@ identifier d_show; identifier dev, attr, buf; identifier len; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... - len += scnprintf(buf + len, PAGE_SIZE - len, + len += sysfs_emit_at(buf, len, ...); ...> return len; } @@ identifier d_show; identifier dev, attr, buf; expression chr; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { ... - strcpy(buf, chr); - return strlen(buf); + return sysfs_emit(buf, chr); } Signed-off-by: Joe Perches <joe@perches.com> Link: https://lore.kernel.org/r/3d033c33056d88bbe34d4ddb62afd05ee166ab9a.1600285923.git.joe@perches.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-09-16 20:40:39 +00:00
return sysfs_emit(buf, "%d\n",
!!(link->flags & DL_FLAG_SYNC_STATE_ONLY));
}
static DEVICE_ATTR_RO(sync_state_only);
static struct attribute *devlink_attrs[] = {
&dev_attr_status.attr,
&dev_attr_auto_remove_on.attr,
&dev_attr_runtime_pm.attr,
&dev_attr_sync_state_only.attr,
NULL,
};
ATTRIBUTE_GROUPS(devlink);
static void device_link_release_fn(struct work_struct *work)
driver core: Fix sleeping in invalid context during device link deletion Marek and Guenter reported that commit 287905e68dd2 ("driver core: Expose device link details in sysfs") caused sleeping/scheduling while atomic warnings. BUG: sleeping function called from invalid context at kernel/locking/mutex.c:935 in_atomic(): 1, irqs_disabled(): 0, non_block: 0, pid: 12, name: kworker/0:1 2 locks held by kworker/0:1/12: #0: ee8074a8 ((wq_completion)rcu_gp){+.+.}-{0:0}, at: process_one_work+0x174/0x7dc #1: ee921f20 ((work_completion)(&sdp->work)){+.+.}-{0:0}, at: process_one_work+0x174/0x7dc Preemption disabled at: [<c01b10f0>] srcu_invoke_callbacks+0xc0/0x154 ----- 8< ----- SNIP [<c064590c>] (device_del) from [<c0645c9c>] (device_unregister+0x24/0x64) [<c0645c9c>] (device_unregister) from [<c01b10fc>] (srcu_invoke_callbacks+0xcc/0x154) [<c01b10fc>] (srcu_invoke_callbacks) from [<c01493c4>] (process_one_work+0x234/0x7dc) [<c01493c4>] (process_one_work) from [<c01499b0>] (worker_thread+0x44/0x51c) [<c01499b0>] (worker_thread) from [<c0150bf4>] (kthread+0x158/0x1a0) [<c0150bf4>] (kthread) from [<c0100114>] (ret_from_fork+0x14/0x20) Exception stack(0xee921fb0 to 0xee921ff8) This was caused by the device link device being released in the context of srcu_invoke_callbacks(). There is no need to wait till the RCU callback to release the device link device. So release the device earlier and move the call_srcu() into the device release code. That way, the memory will get freed only after the device is released AND the RCU callback is called. Fixes: 287905e68dd2 ("driver core: Expose device link details in sysfs") Reported-by: Marek Szyprowski <m.szyprowski@samsung.com> Reported-by: Guenter Roeck <linux@roeck-us.net> Reported-by: Naresh Kamboju <naresh.kamboju@linaro.org> Signed-off-by: Saravana Kannan <saravanak@google.com> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Tested-by: Guenter Roeck <linux@roeck-us.net> Link: https://lore.kernel.org/r/20200716214523.2924704-1-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-07-16 21:45:23 +00:00
{
struct device_link *link = container_of(work, struct device_link, rm_work);
/* Ensure that all references to the link object have been dropped. */
device_link_synchronize_removal();
pm_runtime_release_supplier(link);
PM: runtime: Fix supplier device management during consumer probe Because pm_runtime_get_suppliers() bumps up the rpm_active counter of each device link to a supplier of the given device in addition to bumping up the supplier's PM-runtime usage counter, a runtime suspend of the consumer device may case the latter to go down to 0 when pm_runtime_put_suppliers() is running on a remote CPU. If that happens after pm_runtime_put_suppliers() has released power.lock for the consumer device, and a runtime resume of that device takes place immediately after it, before pm_runtime_put() is called for the supplier, that pm_runtime_put() call may cause the supplier to be suspended even though the consumer is active. To prevent that from happening, modify pm_runtime_get_suppliers() to call pm_runtime_get_sync() for the given device's suppliers without touching the rpm_active counters of the involved device links Accordingly, modify pm_runtime_put_suppliers() to call pm_runtime_put() for the given device's suppliers without looking at the rpm_active counters of the device links at hand. [This is analogous to what happened before commit 4c06c4e6cf63 ("driver core: Fix possible supplier PM-usage counter imbalance").] Since pm_runtime_get_suppliers() sets supplier_preactivated for each device link where the supplier's PM-runtime usage counter has been incremented and pm_runtime_put_suppliers() calls pm_runtime_put() for the suppliers whose device links have supplier_preactivated set, the PM-runtime usage counter is balanced for each supplier and this is independent of the runtime suspend and resume of the consumer device. However, in case a device link with DL_FLAG_PM_RUNTIME set is dropped during the consumer device probe, so pm_runtime_get_suppliers() bumps up the supplier's PM-runtime usage counter, but it cannot be dropped by pm_runtime_put_suppliers(), make device_link_release_fn() take care of that. Fixes: 4c06c4e6cf63 ("driver core: Fix possible supplier PM-usage counter imbalance") Reported-by: Peter Wang <peter.wang@mediatek.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Reviewed-by: Peter Wang <peter.wang@mediatek.com> Cc: 5.1+ <stable@vger.kernel.org> # 5.1+
2022-06-30 19:16:41 +00:00
/*
* If supplier_preactivated is set, the link has been dropped between
* the pm_runtime_get_suppliers() and pm_runtime_put_suppliers() calls
* in __driver_probe_device(). In that case, drop the supplier's
* PM-runtime usage counter to remove the reference taken by
* pm_runtime_get_suppliers().
*/
if (link->supplier_preactivated)
pm_runtime_put_noidle(link->supplier);
pm_request_idle(link->supplier);
driver core: Fix sleeping in invalid context during device link deletion Marek and Guenter reported that commit 287905e68dd2 ("driver core: Expose device link details in sysfs") caused sleeping/scheduling while atomic warnings. BUG: sleeping function called from invalid context at kernel/locking/mutex.c:935 in_atomic(): 1, irqs_disabled(): 0, non_block: 0, pid: 12, name: kworker/0:1 2 locks held by kworker/0:1/12: #0: ee8074a8 ((wq_completion)rcu_gp){+.+.}-{0:0}, at: process_one_work+0x174/0x7dc #1: ee921f20 ((work_completion)(&sdp->work)){+.+.}-{0:0}, at: process_one_work+0x174/0x7dc Preemption disabled at: [<c01b10f0>] srcu_invoke_callbacks+0xc0/0x154 ----- 8< ----- SNIP [<c064590c>] (device_del) from [<c0645c9c>] (device_unregister+0x24/0x64) [<c0645c9c>] (device_unregister) from [<c01b10fc>] (srcu_invoke_callbacks+0xcc/0x154) [<c01b10fc>] (srcu_invoke_callbacks) from [<c01493c4>] (process_one_work+0x234/0x7dc) [<c01493c4>] (process_one_work) from [<c01499b0>] (worker_thread+0x44/0x51c) [<c01499b0>] (worker_thread) from [<c0150bf4>] (kthread+0x158/0x1a0) [<c0150bf4>] (kthread) from [<c0100114>] (ret_from_fork+0x14/0x20) Exception stack(0xee921fb0 to 0xee921ff8) This was caused by the device link device being released in the context of srcu_invoke_callbacks(). There is no need to wait till the RCU callback to release the device link device. So release the device earlier and move the call_srcu() into the device release code. That way, the memory will get freed only after the device is released AND the RCU callback is called. Fixes: 287905e68dd2 ("driver core: Expose device link details in sysfs") Reported-by: Marek Szyprowski <m.szyprowski@samsung.com> Reported-by: Guenter Roeck <linux@roeck-us.net> Reported-by: Naresh Kamboju <naresh.kamboju@linaro.org> Signed-off-by: Saravana Kannan <saravanak@google.com> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Tested-by: Guenter Roeck <linux@roeck-us.net> Link: https://lore.kernel.org/r/20200716214523.2924704-1-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-07-16 21:45:23 +00:00
put_device(link->consumer);
put_device(link->supplier);
kfree(link);
}
static void devlink_dev_release(struct device *dev)
{
driver core: Fix sleeping in invalid context during device link deletion Marek and Guenter reported that commit 287905e68dd2 ("driver core: Expose device link details in sysfs") caused sleeping/scheduling while atomic warnings. BUG: sleeping function called from invalid context at kernel/locking/mutex.c:935 in_atomic(): 1, irqs_disabled(): 0, non_block: 0, pid: 12, name: kworker/0:1 2 locks held by kworker/0:1/12: #0: ee8074a8 ((wq_completion)rcu_gp){+.+.}-{0:0}, at: process_one_work+0x174/0x7dc #1: ee921f20 ((work_completion)(&sdp->work)){+.+.}-{0:0}, at: process_one_work+0x174/0x7dc Preemption disabled at: [<c01b10f0>] srcu_invoke_callbacks+0xc0/0x154 ----- 8< ----- SNIP [<c064590c>] (device_del) from [<c0645c9c>] (device_unregister+0x24/0x64) [<c0645c9c>] (device_unregister) from [<c01b10fc>] (srcu_invoke_callbacks+0xcc/0x154) [<c01b10fc>] (srcu_invoke_callbacks) from [<c01493c4>] (process_one_work+0x234/0x7dc) [<c01493c4>] (process_one_work) from [<c01499b0>] (worker_thread+0x44/0x51c) [<c01499b0>] (worker_thread) from [<c0150bf4>] (kthread+0x158/0x1a0) [<c0150bf4>] (kthread) from [<c0100114>] (ret_from_fork+0x14/0x20) Exception stack(0xee921fb0 to 0xee921ff8) This was caused by the device link device being released in the context of srcu_invoke_callbacks(). There is no need to wait till the RCU callback to release the device link device. So release the device earlier and move the call_srcu() into the device release code. That way, the memory will get freed only after the device is released AND the RCU callback is called. Fixes: 287905e68dd2 ("driver core: Expose device link details in sysfs") Reported-by: Marek Szyprowski <m.szyprowski@samsung.com> Reported-by: Guenter Roeck <linux@roeck-us.net> Reported-by: Naresh Kamboju <naresh.kamboju@linaro.org> Signed-off-by: Saravana Kannan <saravanak@google.com> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Tested-by: Guenter Roeck <linux@roeck-us.net> Link: https://lore.kernel.org/r/20200716214523.2924704-1-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-07-16 21:45:23 +00:00
struct device_link *link = to_devlink(dev);
INIT_WORK(&link->rm_work, device_link_release_fn);
/*
* It may take a while to complete this work because of the SRCU
* synchronization in device_link_release_fn() and if the consumer or
* supplier devices get deleted when it runs, so put it into the "long"
* workqueue.
*/
queue_work(system_long_wq, &link->rm_work);
driver core: Fix sleeping in invalid context during device link deletion Marek and Guenter reported that commit 287905e68dd2 ("driver core: Expose device link details in sysfs") caused sleeping/scheduling while atomic warnings. BUG: sleeping function called from invalid context at kernel/locking/mutex.c:935 in_atomic(): 1, irqs_disabled(): 0, non_block: 0, pid: 12, name: kworker/0:1 2 locks held by kworker/0:1/12: #0: ee8074a8 ((wq_completion)rcu_gp){+.+.}-{0:0}, at: process_one_work+0x174/0x7dc #1: ee921f20 ((work_completion)(&sdp->work)){+.+.}-{0:0}, at: process_one_work+0x174/0x7dc Preemption disabled at: [<c01b10f0>] srcu_invoke_callbacks+0xc0/0x154 ----- 8< ----- SNIP [<c064590c>] (device_del) from [<c0645c9c>] (device_unregister+0x24/0x64) [<c0645c9c>] (device_unregister) from [<c01b10fc>] (srcu_invoke_callbacks+0xcc/0x154) [<c01b10fc>] (srcu_invoke_callbacks) from [<c01493c4>] (process_one_work+0x234/0x7dc) [<c01493c4>] (process_one_work) from [<c01499b0>] (worker_thread+0x44/0x51c) [<c01499b0>] (worker_thread) from [<c0150bf4>] (kthread+0x158/0x1a0) [<c0150bf4>] (kthread) from [<c0100114>] (ret_from_fork+0x14/0x20) Exception stack(0xee921fb0 to 0xee921ff8) This was caused by the device link device being released in the context of srcu_invoke_callbacks(). There is no need to wait till the RCU callback to release the device link device. So release the device earlier and move the call_srcu() into the device release code. That way, the memory will get freed only after the device is released AND the RCU callback is called. Fixes: 287905e68dd2 ("driver core: Expose device link details in sysfs") Reported-by: Marek Szyprowski <m.szyprowski@samsung.com> Reported-by: Guenter Roeck <linux@roeck-us.net> Reported-by: Naresh Kamboju <naresh.kamboju@linaro.org> Signed-off-by: Saravana Kannan <saravanak@google.com> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Tested-by: Guenter Roeck <linux@roeck-us.net> Link: https://lore.kernel.org/r/20200716214523.2924704-1-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-07-16 21:45:23 +00:00
}
static struct class devlink_class = {
.name = "devlink",
.owner = THIS_MODULE,
.dev_groups = devlink_groups,
.dev_release = devlink_dev_release,
};
static int devlink_add_symlinks(struct device *dev,
struct class_interface *class_intf)
{
int ret;
size_t len;
struct device_link *link = to_devlink(dev);
struct device *sup = link->supplier;
struct device *con = link->consumer;
char *buf;
len = max(strlen(dev_bus_name(sup)) + strlen(dev_name(sup)),
strlen(dev_bus_name(con)) + strlen(dev_name(con)));
len += strlen(":");
len += strlen("supplier:") + 1;
buf = kzalloc(len, GFP_KERNEL);
if (!buf)
return -ENOMEM;
ret = sysfs_create_link(&link->link_dev.kobj, &sup->kobj, "supplier");
if (ret)
goto out;
ret = sysfs_create_link(&link->link_dev.kobj, &con->kobj, "consumer");
if (ret)
goto err_con;
snprintf(buf, len, "consumer:%s:%s", dev_bus_name(con), dev_name(con));
ret = sysfs_create_link(&sup->kobj, &link->link_dev.kobj, buf);
if (ret)
goto err_con_dev;
snprintf(buf, len, "supplier:%s:%s", dev_bus_name(sup), dev_name(sup));
ret = sysfs_create_link(&con->kobj, &link->link_dev.kobj, buf);
if (ret)
goto err_sup_dev;
goto out;
err_sup_dev:
snprintf(buf, len, "consumer:%s:%s", dev_bus_name(con), dev_name(con));
sysfs_remove_link(&sup->kobj, buf);
err_con_dev:
sysfs_remove_link(&link->link_dev.kobj, "consumer");
err_con:
sysfs_remove_link(&link->link_dev.kobj, "supplier");
out:
kfree(buf);
return ret;
}
static void devlink_remove_symlinks(struct device *dev,
struct class_interface *class_intf)
{
struct device_link *link = to_devlink(dev);
size_t len;
struct device *sup = link->supplier;
struct device *con = link->consumer;
char *buf;
sysfs_remove_link(&link->link_dev.kobj, "consumer");
sysfs_remove_link(&link->link_dev.kobj, "supplier");
len = max(strlen(dev_bus_name(sup)) + strlen(dev_name(sup)),
strlen(dev_bus_name(con)) + strlen(dev_name(con)));
len += strlen(":");
len += strlen("supplier:") + 1;
buf = kzalloc(len, GFP_KERNEL);
if (!buf) {
WARN(1, "Unable to properly free device link symlinks!\n");
return;
}
if (device_is_registered(con)) {
snprintf(buf, len, "supplier:%s:%s", dev_bus_name(sup), dev_name(sup));
sysfs_remove_link(&con->kobj, buf);
}
snprintf(buf, len, "consumer:%s:%s", dev_bus_name(con), dev_name(con));
sysfs_remove_link(&sup->kobj, buf);
kfree(buf);
}
static struct class_interface devlink_class_intf = {
.class = &devlink_class,
.add_dev = devlink_add_symlinks,
.remove_dev = devlink_remove_symlinks,
};
static int __init devlink_class_init(void)
{
int ret;
ret = class_register(&devlink_class);
if (ret)
return ret;
ret = class_interface_register(&devlink_class_intf);
if (ret)
class_unregister(&devlink_class);
return ret;
}
postcore_initcall(devlink_class_init);
driver core: Remove device link creation limitation If device_link_add() is called for a consumer/supplier pair with an existing device link between them and the existing link's type is not in agreement with the flags passed to that function by its caller, NULL will be returned. That is seriously inconvenient, because it forces the callers of device_link_add() to worry about what others may or may not do even if that is not relevant to them for any other reasons. It turns out, however, that this limitation can be made go away relatively easily. The underlying observation is that if DL_FLAG_STATELESS has been passed to device_link_add() in flags for the given consumer/supplier pair at least once, calling either device_link_del() or device_link_remove() to release the link returned by it should work, but there are no other requirements associated with that flag. In turn, if at least one of the callers of device_link_add() for the given consumer/supplier pair has not passed DL_FLAG_STATELESS to it in flags, the driver core should track the status of the link and act on it as appropriate (ie. the link should be treated as "managed"). This means that DL_FLAG_STATELESS needs to be set for managed device links and it should be valid to call device_link_del() or device_link_remove() to drop references to them in certain sutiations. To allow that to happen, introduce a new (internal) device link flag called DL_FLAG_MANAGED and make device_link_add() set it automatically whenever DL_FLAG_STATELESS is not passed to it. Also make it take additional references to existing device links that were previously stateless (that is, with DL_FLAG_STATELESS set and DL_FLAG_MANAGED unset) and will need to be managed going forward and initialize their status (which has been DL_STATE_NONE so far). Accordingly, when a managed device link is dropped automatically by the driver core, make it clear DL_FLAG_MANAGED, reset the link's status back to DL_STATE_NONE and drop the reference to it associated with DL_FLAG_MANAGED instead of just deleting it right away (to allow it to stay around in case it still needs to be released explicitly by someone). With that, since setting DL_FLAG_STATELESS doesn't mean that the device link in question is not managed any more, replace all of the status-tracking checks against DL_FLAG_STATELESS with analogous checks against DL_FLAG_MANAGED and update the documentation to reflect these changes. While at it, make device_link_add() reject flags that it does not recognize, including DL_FLAG_MANAGED. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Reviewed-by: Saravana Kannan <saravanak@google.com> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Review-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/2305283.AStDPdUUnE@kreacher Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-07-16 15:21:06 +00:00
#define DL_MANAGED_LINK_FLAGS (DL_FLAG_AUTOREMOVE_CONSUMER | \
DL_FLAG_AUTOREMOVE_SUPPLIER | \
DL_FLAG_AUTOPROBE_CONSUMER | \
DL_FLAG_SYNC_STATE_ONLY | \
DL_FLAG_INFERRED)
driver core: Remove device link creation limitation If device_link_add() is called for a consumer/supplier pair with an existing device link between them and the existing link's type is not in agreement with the flags passed to that function by its caller, NULL will be returned. That is seriously inconvenient, because it forces the callers of device_link_add() to worry about what others may or may not do even if that is not relevant to them for any other reasons. It turns out, however, that this limitation can be made go away relatively easily. The underlying observation is that if DL_FLAG_STATELESS has been passed to device_link_add() in flags for the given consumer/supplier pair at least once, calling either device_link_del() or device_link_remove() to release the link returned by it should work, but there are no other requirements associated with that flag. In turn, if at least one of the callers of device_link_add() for the given consumer/supplier pair has not passed DL_FLAG_STATELESS to it in flags, the driver core should track the status of the link and act on it as appropriate (ie. the link should be treated as "managed"). This means that DL_FLAG_STATELESS needs to be set for managed device links and it should be valid to call device_link_del() or device_link_remove() to drop references to them in certain sutiations. To allow that to happen, introduce a new (internal) device link flag called DL_FLAG_MANAGED and make device_link_add() set it automatically whenever DL_FLAG_STATELESS is not passed to it. Also make it take additional references to existing device links that were previously stateless (that is, with DL_FLAG_STATELESS set and DL_FLAG_MANAGED unset) and will need to be managed going forward and initialize their status (which has been DL_STATE_NONE so far). Accordingly, when a managed device link is dropped automatically by the driver core, make it clear DL_FLAG_MANAGED, reset the link's status back to DL_STATE_NONE and drop the reference to it associated with DL_FLAG_MANAGED instead of just deleting it right away (to allow it to stay around in case it still needs to be released explicitly by someone). With that, since setting DL_FLAG_STATELESS doesn't mean that the device link in question is not managed any more, replace all of the status-tracking checks against DL_FLAG_STATELESS with analogous checks against DL_FLAG_MANAGED and update the documentation to reflect these changes. While at it, make device_link_add() reject flags that it does not recognize, including DL_FLAG_MANAGED. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Reviewed-by: Saravana Kannan <saravanak@google.com> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Review-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/2305283.AStDPdUUnE@kreacher Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-07-16 15:21:06 +00:00
#define DL_ADD_VALID_FLAGS (DL_MANAGED_LINK_FLAGS | DL_FLAG_STATELESS | \
DL_FLAG_PM_RUNTIME | DL_FLAG_RPM_ACTIVE)
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
/**
* device_link_add - Create a link between two devices.
* @consumer: Consumer end of the link.
* @supplier: Supplier end of the link.
* @flags: Link flags.
*
* The caller is responsible for the proper synchronization of the link creation
* with runtime PM. First, setting the DL_FLAG_PM_RUNTIME flag will cause the
* runtime PM framework to take the link into account. Second, if the
* DL_FLAG_RPM_ACTIVE flag is set in addition to it, the supplier devices will
* be forced into the active meta state and reference-counted upon the creation
* of the link. If DL_FLAG_PM_RUNTIME is not set, DL_FLAG_RPM_ACTIVE will be
* ignored.
*
driver core: Remove device link creation limitation If device_link_add() is called for a consumer/supplier pair with an existing device link between them and the existing link's type is not in agreement with the flags passed to that function by its caller, NULL will be returned. That is seriously inconvenient, because it forces the callers of device_link_add() to worry about what others may or may not do even if that is not relevant to them for any other reasons. It turns out, however, that this limitation can be made go away relatively easily. The underlying observation is that if DL_FLAG_STATELESS has been passed to device_link_add() in flags for the given consumer/supplier pair at least once, calling either device_link_del() or device_link_remove() to release the link returned by it should work, but there are no other requirements associated with that flag. In turn, if at least one of the callers of device_link_add() for the given consumer/supplier pair has not passed DL_FLAG_STATELESS to it in flags, the driver core should track the status of the link and act on it as appropriate (ie. the link should be treated as "managed"). This means that DL_FLAG_STATELESS needs to be set for managed device links and it should be valid to call device_link_del() or device_link_remove() to drop references to them in certain sutiations. To allow that to happen, introduce a new (internal) device link flag called DL_FLAG_MANAGED and make device_link_add() set it automatically whenever DL_FLAG_STATELESS is not passed to it. Also make it take additional references to existing device links that were previously stateless (that is, with DL_FLAG_STATELESS set and DL_FLAG_MANAGED unset) and will need to be managed going forward and initialize their status (which has been DL_STATE_NONE so far). Accordingly, when a managed device link is dropped automatically by the driver core, make it clear DL_FLAG_MANAGED, reset the link's status back to DL_STATE_NONE and drop the reference to it associated with DL_FLAG_MANAGED instead of just deleting it right away (to allow it to stay around in case it still needs to be released explicitly by someone). With that, since setting DL_FLAG_STATELESS doesn't mean that the device link in question is not managed any more, replace all of the status-tracking checks against DL_FLAG_STATELESS with analogous checks against DL_FLAG_MANAGED and update the documentation to reflect these changes. While at it, make device_link_add() reject flags that it does not recognize, including DL_FLAG_MANAGED. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Reviewed-by: Saravana Kannan <saravanak@google.com> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Review-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/2305283.AStDPdUUnE@kreacher Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-07-16 15:21:06 +00:00
* If DL_FLAG_STATELESS is set in @flags, the caller of this function is
* expected to release the link returned by it directly with the help of either
* device_link_del() or device_link_remove().
driver core: Make driver core own stateful device links Even though stateful device links are managed by the driver core in principle, their creators are allowed and sometimes even expected to drop references to them via device_link_del() or device_link_remove(), but that doesn't really play well with the "persistent" link concept. If "persistent" managed device links are created from driver probe callbacks, device_link_add() called to do that will take a new reference on the link each time the callback runs and those references will never be dropped, which kind of isn't nice. This issues arises because of the link reference counting carried out by device_link_add() for existing links, but that is only done to avoid deleting device links that may still be necessary, which shouldn't be a concern for managed (stateful) links. These device links are managed by the driver core and whoever creates one of them will need it at least as long as until the consumer driver is detached from its device and deleting it may be left to the driver core just fine. For this reason, rework device_link_add() to apply the reference counting to stateless links only and make device_link_del() and device_link_remove() drop references to stateless links only too. After this change, if called to add a stateful device link for a consumer-supplier pair for which a stateful device link is present already, device_link_add() will return the existing link without incrementing its reference counter. Accordingly, device_link_del() and device_link_remove() will WARN() and do nothing when called to drop a reference to a stateful link. Thus, effectively, all stateful device links will be owned by the driver core. In addition, clean up the handling of the link management flags, DL_FLAG_AUTOREMOVE_CONSUMER and DL_FLAG_AUTOREMOVE_SUPPLIER, so that (a) they are never set at the same time and (b) if device_link_add() is called for a consumer-supplier pair with an existing stateful link between them, the flags of that link will be combined with the flags passed to device_link_add() to ensure that the life time of the link is sufficient for all of the callers of device_link_add() for the same consumer-supplier pair. Update the device_link_add() kerneldoc comment to reflect the above changes. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-02-01 00:58:33 +00:00
*
* If that flag is not set, however, the caller of this function is handing the
* management of the link over to the driver core entirely and its return value
* can only be used to check whether or not the link is present. In that case,
* the DL_FLAG_AUTOREMOVE_CONSUMER and DL_FLAG_AUTOREMOVE_SUPPLIER device link
* flags can be used to indicate to the driver core when the link can be safely
* deleted. Namely, setting one of them in @flags indicates to the driver core
* that the link is not going to be used (by the given caller of this function)
* after unbinding the consumer or supplier driver, respectively, from its
* device, so the link can be deleted at that point. If none of them is set,
* the link will be maintained until one of the devices pointed to by it (either
* the consumer or the supplier) is unregistered.
*
* Also, if DL_FLAG_STATELESS, DL_FLAG_AUTOREMOVE_CONSUMER and
* DL_FLAG_AUTOREMOVE_SUPPLIER are not set in @flags (that is, a persistent
* managed device link is being added), the DL_FLAG_AUTOPROBE_CONSUMER flag can
* be used to request the driver core to automatically probe for a consumer
* driver after successfully binding a driver to the supplier device.
*
driver core: Remove device link creation limitation If device_link_add() is called for a consumer/supplier pair with an existing device link between them and the existing link's type is not in agreement with the flags passed to that function by its caller, NULL will be returned. That is seriously inconvenient, because it forces the callers of device_link_add() to worry about what others may or may not do even if that is not relevant to them for any other reasons. It turns out, however, that this limitation can be made go away relatively easily. The underlying observation is that if DL_FLAG_STATELESS has been passed to device_link_add() in flags for the given consumer/supplier pair at least once, calling either device_link_del() or device_link_remove() to release the link returned by it should work, but there are no other requirements associated with that flag. In turn, if at least one of the callers of device_link_add() for the given consumer/supplier pair has not passed DL_FLAG_STATELESS to it in flags, the driver core should track the status of the link and act on it as appropriate (ie. the link should be treated as "managed"). This means that DL_FLAG_STATELESS needs to be set for managed device links and it should be valid to call device_link_del() or device_link_remove() to drop references to them in certain sutiations. To allow that to happen, introduce a new (internal) device link flag called DL_FLAG_MANAGED and make device_link_add() set it automatically whenever DL_FLAG_STATELESS is not passed to it. Also make it take additional references to existing device links that were previously stateless (that is, with DL_FLAG_STATELESS set and DL_FLAG_MANAGED unset) and will need to be managed going forward and initialize their status (which has been DL_STATE_NONE so far). Accordingly, when a managed device link is dropped automatically by the driver core, make it clear DL_FLAG_MANAGED, reset the link's status back to DL_STATE_NONE and drop the reference to it associated with DL_FLAG_MANAGED instead of just deleting it right away (to allow it to stay around in case it still needs to be released explicitly by someone). With that, since setting DL_FLAG_STATELESS doesn't mean that the device link in question is not managed any more, replace all of the status-tracking checks against DL_FLAG_STATELESS with analogous checks against DL_FLAG_MANAGED and update the documentation to reflect these changes. While at it, make device_link_add() reject flags that it does not recognize, including DL_FLAG_MANAGED. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Reviewed-by: Saravana Kannan <saravanak@google.com> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Review-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/2305283.AStDPdUUnE@kreacher Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-07-16 15:21:06 +00:00
* The combination of DL_FLAG_STATELESS and one of DL_FLAG_AUTOREMOVE_CONSUMER,
* DL_FLAG_AUTOREMOVE_SUPPLIER, or DL_FLAG_AUTOPROBE_CONSUMER set in @flags at
* the same time is invalid and will cause NULL to be returned upfront.
* However, if a device link between the given @consumer and @supplier pair
* exists already when this function is called for them, the existing link will
* be returned regardless of its current type and status (the link's flags may
* be modified then). The caller of this function is then expected to treat
* the link as though it has just been created, so (in particular) if
* DL_FLAG_STATELESS was passed in @flags, the link needs to be released
* explicitly when not needed any more (as stated above).
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
*
* A side effect of the link creation is re-ordering of dpm_list and the
* devices_kset list by moving the consumer device and all devices depending
* on it to the ends of these lists (that does not happen to devices that have
* not been registered when this function is called).
*
* The supplier device is required to be registered when this function is called
* and NULL will be returned if that is not the case. The consumer device need
* not be registered, however.
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
*/
struct device_link *device_link_add(struct device *consumer,
struct device *supplier, u32 flags)
{
struct device_link *link;
if (!consumer || !supplier || consumer == supplier ||
flags & ~DL_ADD_VALID_FLAGS ||
driver core: Remove device link creation limitation If device_link_add() is called for a consumer/supplier pair with an existing device link between them and the existing link's type is not in agreement with the flags passed to that function by its caller, NULL will be returned. That is seriously inconvenient, because it forces the callers of device_link_add() to worry about what others may or may not do even if that is not relevant to them for any other reasons. It turns out, however, that this limitation can be made go away relatively easily. The underlying observation is that if DL_FLAG_STATELESS has been passed to device_link_add() in flags for the given consumer/supplier pair at least once, calling either device_link_del() or device_link_remove() to release the link returned by it should work, but there are no other requirements associated with that flag. In turn, if at least one of the callers of device_link_add() for the given consumer/supplier pair has not passed DL_FLAG_STATELESS to it in flags, the driver core should track the status of the link and act on it as appropriate (ie. the link should be treated as "managed"). This means that DL_FLAG_STATELESS needs to be set for managed device links and it should be valid to call device_link_del() or device_link_remove() to drop references to them in certain sutiations. To allow that to happen, introduce a new (internal) device link flag called DL_FLAG_MANAGED and make device_link_add() set it automatically whenever DL_FLAG_STATELESS is not passed to it. Also make it take additional references to existing device links that were previously stateless (that is, with DL_FLAG_STATELESS set and DL_FLAG_MANAGED unset) and will need to be managed going forward and initialize their status (which has been DL_STATE_NONE so far). Accordingly, when a managed device link is dropped automatically by the driver core, make it clear DL_FLAG_MANAGED, reset the link's status back to DL_STATE_NONE and drop the reference to it associated with DL_FLAG_MANAGED instead of just deleting it right away (to allow it to stay around in case it still needs to be released explicitly by someone). With that, since setting DL_FLAG_STATELESS doesn't mean that the device link in question is not managed any more, replace all of the status-tracking checks against DL_FLAG_STATELESS with analogous checks against DL_FLAG_MANAGED and update the documentation to reflect these changes. While at it, make device_link_add() reject flags that it does not recognize, including DL_FLAG_MANAGED. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Reviewed-by: Saravana Kannan <saravanak@google.com> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Review-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/2305283.AStDPdUUnE@kreacher Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-07-16 15:21:06 +00:00
(flags & DL_FLAG_STATELESS && flags & DL_MANAGED_LINK_FLAGS) ||
(flags & DL_FLAG_SYNC_STATE_ONLY &&
(flags & ~DL_FLAG_INFERRED) != DL_FLAG_SYNC_STATE_ONLY) ||
(flags & DL_FLAG_AUTOPROBE_CONSUMER &&
flags & (DL_FLAG_AUTOREMOVE_CONSUMER |
DL_FLAG_AUTOREMOVE_SUPPLIER)))
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
return NULL;
if (flags & DL_FLAG_PM_RUNTIME && flags & DL_FLAG_RPM_ACTIVE) {
if (pm_runtime_get_sync(supplier) < 0) {
pm_runtime_put_noidle(supplier);
return NULL;
}
}
driver core: Remove device link creation limitation If device_link_add() is called for a consumer/supplier pair with an existing device link between them and the existing link's type is not in agreement with the flags passed to that function by its caller, NULL will be returned. That is seriously inconvenient, because it forces the callers of device_link_add() to worry about what others may or may not do even if that is not relevant to them for any other reasons. It turns out, however, that this limitation can be made go away relatively easily. The underlying observation is that if DL_FLAG_STATELESS has been passed to device_link_add() in flags for the given consumer/supplier pair at least once, calling either device_link_del() or device_link_remove() to release the link returned by it should work, but there are no other requirements associated with that flag. In turn, if at least one of the callers of device_link_add() for the given consumer/supplier pair has not passed DL_FLAG_STATELESS to it in flags, the driver core should track the status of the link and act on it as appropriate (ie. the link should be treated as "managed"). This means that DL_FLAG_STATELESS needs to be set for managed device links and it should be valid to call device_link_del() or device_link_remove() to drop references to them in certain sutiations. To allow that to happen, introduce a new (internal) device link flag called DL_FLAG_MANAGED and make device_link_add() set it automatically whenever DL_FLAG_STATELESS is not passed to it. Also make it take additional references to existing device links that were previously stateless (that is, with DL_FLAG_STATELESS set and DL_FLAG_MANAGED unset) and will need to be managed going forward and initialize their status (which has been DL_STATE_NONE so far). Accordingly, when a managed device link is dropped automatically by the driver core, make it clear DL_FLAG_MANAGED, reset the link's status back to DL_STATE_NONE and drop the reference to it associated with DL_FLAG_MANAGED instead of just deleting it right away (to allow it to stay around in case it still needs to be released explicitly by someone). With that, since setting DL_FLAG_STATELESS doesn't mean that the device link in question is not managed any more, replace all of the status-tracking checks against DL_FLAG_STATELESS with analogous checks against DL_FLAG_MANAGED and update the documentation to reflect these changes. While at it, make device_link_add() reject flags that it does not recognize, including DL_FLAG_MANAGED. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Reviewed-by: Saravana Kannan <saravanak@google.com> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Review-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/2305283.AStDPdUUnE@kreacher Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-07-16 15:21:06 +00:00
if (!(flags & DL_FLAG_STATELESS))
flags |= DL_FLAG_MANAGED;
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
device_links_write_lock();
device_pm_lock();
/*
* If the supplier has not been fully registered yet or there is a
* reverse (non-SYNC_STATE_ONLY) dependency between the consumer and
* the supplier already in the graph, return NULL. If the link is a
* SYNC_STATE_ONLY link, we don't check for reverse dependencies
* because it only affects sync_state() callbacks.
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
*/
if (!device_pm_initialized(supplier)
|| (!(flags & DL_FLAG_SYNC_STATE_ONLY) &&
device_is_dependent(consumer, supplier))) {
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
link = NULL;
goto out;
}
/*
* SYNC_STATE_ONLY links are useless once a consumer device has probed.
* So, only create it if the consumer hasn't probed yet.
*/
if (flags & DL_FLAG_SYNC_STATE_ONLY &&
consumer->links.status != DL_DEV_NO_DRIVER &&
consumer->links.status != DL_DEV_PROBING) {
link = NULL;
goto out;
}
driver core: Make driver core own stateful device links Even though stateful device links are managed by the driver core in principle, their creators are allowed and sometimes even expected to drop references to them via device_link_del() or device_link_remove(), but that doesn't really play well with the "persistent" link concept. If "persistent" managed device links are created from driver probe callbacks, device_link_add() called to do that will take a new reference on the link each time the callback runs and those references will never be dropped, which kind of isn't nice. This issues arises because of the link reference counting carried out by device_link_add() for existing links, but that is only done to avoid deleting device links that may still be necessary, which shouldn't be a concern for managed (stateful) links. These device links are managed by the driver core and whoever creates one of them will need it at least as long as until the consumer driver is detached from its device and deleting it may be left to the driver core just fine. For this reason, rework device_link_add() to apply the reference counting to stateless links only and make device_link_del() and device_link_remove() drop references to stateless links only too. After this change, if called to add a stateful device link for a consumer-supplier pair for which a stateful device link is present already, device_link_add() will return the existing link without incrementing its reference counter. Accordingly, device_link_del() and device_link_remove() will WARN() and do nothing when called to drop a reference to a stateful link. Thus, effectively, all stateful device links will be owned by the driver core. In addition, clean up the handling of the link management flags, DL_FLAG_AUTOREMOVE_CONSUMER and DL_FLAG_AUTOREMOVE_SUPPLIER, so that (a) they are never set at the same time and (b) if device_link_add() is called for a consumer-supplier pair with an existing stateful link between them, the flags of that link will be combined with the flags passed to device_link_add() to ensure that the life time of the link is sufficient for all of the callers of device_link_add() for the same consumer-supplier pair. Update the device_link_add() kerneldoc comment to reflect the above changes. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-02-01 00:58:33 +00:00
/*
* DL_FLAG_AUTOREMOVE_SUPPLIER indicates that the link will be needed
* longer than for DL_FLAG_AUTOREMOVE_CONSUMER and setting them both
* together doesn't make sense, so prefer DL_FLAG_AUTOREMOVE_SUPPLIER.
*/
if (flags & DL_FLAG_AUTOREMOVE_SUPPLIER)
flags &= ~DL_FLAG_AUTOREMOVE_CONSUMER;
list_for_each_entry(link, &supplier->links.consumers, s_node) {
if (link->consumer != consumer)
continue;
if (link->flags & DL_FLAG_INFERRED &&
!(flags & DL_FLAG_INFERRED))
link->flags &= ~DL_FLAG_INFERRED;
driver core: Fix handling of runtime PM flags in device_link_add() After commit ead18c23c263 ("driver core: Introduce device links reference counting"), if there is a link between the given supplier and the given consumer already, device_link_add() will refcount it and return it unconditionally without updating its flags. It is possible, however, that the second (or any subsequent) caller of device_link_add() for the same consumer-supplier pair will pass DL_FLAG_PM_RUNTIME, possibly along with DL_FLAG_RPM_ACTIVE, in flags to it and the existing link may not behave as expected then. First, if DL_FLAG_PM_RUNTIME is not set in the existing link's flags at all, it needs to be set like during the original initialization of the link. Second, if DL_FLAG_RPM_ACTIVE is passed to device_link_add() in flags (in addition to DL_FLAG_PM_RUNTIME), the existing link should to be updated to reflect the "active" runtime PM configuration of the consumer-supplier pair and extra care must be taken here to avoid possible destructive races with runtime PM of the consumer. To that end, redefine the rpm_active field in struct device_link as a refcount, initialize it to 1 and make rpm_resume() (for the consumer) and device_link_add() increment it whenever they acquire a runtime PM reference on the supplier device. Accordingly, make rpm_suspend() (for the consumer) and pm_runtime_clean_up_links() decrement it and drop runtime PM references to the supplier device in a loop until rpm_active becones 1 again. Fixes: ead18c23c263 ("driver core: Introduce device links reference counting") Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-02-01 00:49:14 +00:00
if (flags & DL_FLAG_PM_RUNTIME) {
if (!(link->flags & DL_FLAG_PM_RUNTIME)) {
driver core: Fix possible supplier PM-usage counter imbalance If a stateless device link to a certain supplier with DL_FLAG_PM_RUNTIME set in the flags is added and then removed by the consumer driver's probe callback, the supplier's PM-runtime usage counter will be nonzero after that which effectively causes the supplier to remain "always on" going forward. Namely, device_link_add() called to add the link invokes device_link_rpm_prepare() which notices that the consumer driver is probing, so it increments the supplier's PM-runtime usage counter with the assumption that the link will stay around until pm_runtime_put_suppliers() is called by driver_probe_device(), but if the link goes away before that point, the supplier's PM-runtime usage counter will remain nonzero. To prevent that from happening, first rework pm_runtime_get_suppliers() and pm_runtime_put_suppliers() to use the rpm_active refounts of device links and make the latter only drop rpm_active and the supplier's PM-runtime usage counter for each link by one, unless rpm_active is one already for it. Next, modify device_link_add() to bump up the new link's rpm_active refcount and the suppliers PM-runtime usage counter by two, to prevent pm_runtime_put_suppliers(), if it is called subsequently, from suspending the supplier prematurely (in case its PM-runtime usage counter goes down to 0 in there). Due to the way rpm_put_suppliers() works, this change does not affect runtime suspend of the consumer ends of new device links (or, generally, device links for which DL_FLAG_PM_RUNTIME has just been set). Fixes: e2f3cd831a28 ("driver core: Fix handling of runtime PM flags in device_link_add()") Reported-by: Ulf Hansson <ulf.hansson@linaro.org> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org> Tested-by: Ulf Hansson <ulf.hansson@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-02-12 12:08:10 +00:00
pm_runtime_new_link(consumer);
driver core: Fix handling of runtime PM flags in device_link_add() After commit ead18c23c263 ("driver core: Introduce device links reference counting"), if there is a link between the given supplier and the given consumer already, device_link_add() will refcount it and return it unconditionally without updating its flags. It is possible, however, that the second (or any subsequent) caller of device_link_add() for the same consumer-supplier pair will pass DL_FLAG_PM_RUNTIME, possibly along with DL_FLAG_RPM_ACTIVE, in flags to it and the existing link may not behave as expected then. First, if DL_FLAG_PM_RUNTIME is not set in the existing link's flags at all, it needs to be set like during the original initialization of the link. Second, if DL_FLAG_RPM_ACTIVE is passed to device_link_add() in flags (in addition to DL_FLAG_PM_RUNTIME), the existing link should to be updated to reflect the "active" runtime PM configuration of the consumer-supplier pair and extra care must be taken here to avoid possible destructive races with runtime PM of the consumer. To that end, redefine the rpm_active field in struct device_link as a refcount, initialize it to 1 and make rpm_resume() (for the consumer) and device_link_add() increment it whenever they acquire a runtime PM reference on the supplier device. Accordingly, make rpm_suspend() (for the consumer) and pm_runtime_clean_up_links() decrement it and drop runtime PM references to the supplier device in a loop until rpm_active becones 1 again. Fixes: ead18c23c263 ("driver core: Introduce device links reference counting") Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-02-01 00:49:14 +00:00
link->flags |= DL_FLAG_PM_RUNTIME;
}
if (flags & DL_FLAG_RPM_ACTIVE)
driver core: Fix PM-runtime for links added during consumer probe Commit 4c06c4e6cf63 ("driver core: Fix possible supplier PM-usage counter imbalance") introduced a regression that causes suppliers to be suspended prematurely for device links added during consumer driver probe if the initial PM-runtime status of the consumer is "suspended" and the consumer is resumed after adding the link and before pm_runtime_put_suppliers() is called. In that case, pm_runtime_put_suppliers() will drop the rpm_active refcount for the link by one and (since rpm_active is equal to two after the preceding consumer resume) the supplier's PM-runtime usage counter will be decremented, which may cause the supplier to suspend even though the consumer's PM-runtime status is "active". For this reason, partially revert commit 4c06c4e6cf63 as the problem it tried to fix needs to be addressed somewhat differently, and change pm_runtime_get_suppliers() and pm_runtime_put_suppliers() so that the latter only drops rpm_active references acquired by the former. [This requires adding a new field to struct device_link, but I coulnd't find a cleaner way to address the issue that would work in all cases.] This causes pm_runtime_put_suppliers() to effectively ignore device links added during consumer probe, so device_link_add() doesn't need to worry about ensuring that suppliers will remain active after pm_runtime_put_suppliers() for links created with DL_FLAG_RPM_ACTIVE set and it only needs to bump up rpm_active by one for those links, so pm_runtime_active_link() is not necessary any more. Fixes: 4c06c4e6cf63 ("driver core: Fix possible supplier PM-usage counter imbalance") Reported-by: Jon Hunter <jonathanh@nvidia.com> Tested-by: Jon Hunter <jonathanh@nvidia.com> Tested-by: Ulf Hansson <ulf.hansson@linaro.org> Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Tested-by: Thierry Reding <treding@nvidia.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-02-19 16:53:26 +00:00
refcount_inc(&link->rpm_active);
driver core: Fix handling of runtime PM flags in device_link_add() After commit ead18c23c263 ("driver core: Introduce device links reference counting"), if there is a link between the given supplier and the given consumer already, device_link_add() will refcount it and return it unconditionally without updating its flags. It is possible, however, that the second (or any subsequent) caller of device_link_add() for the same consumer-supplier pair will pass DL_FLAG_PM_RUNTIME, possibly along with DL_FLAG_RPM_ACTIVE, in flags to it and the existing link may not behave as expected then. First, if DL_FLAG_PM_RUNTIME is not set in the existing link's flags at all, it needs to be set like during the original initialization of the link. Second, if DL_FLAG_RPM_ACTIVE is passed to device_link_add() in flags (in addition to DL_FLAG_PM_RUNTIME), the existing link should to be updated to reflect the "active" runtime PM configuration of the consumer-supplier pair and extra care must be taken here to avoid possible destructive races with runtime PM of the consumer. To that end, redefine the rpm_active field in struct device_link as a refcount, initialize it to 1 and make rpm_resume() (for the consumer) and device_link_add() increment it whenever they acquire a runtime PM reference on the supplier device. Accordingly, make rpm_suspend() (for the consumer) and pm_runtime_clean_up_links() decrement it and drop runtime PM references to the supplier device in a loop until rpm_active becones 1 again. Fixes: ead18c23c263 ("driver core: Introduce device links reference counting") Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-02-01 00:49:14 +00:00
}
driver core: Make driver core own stateful device links Even though stateful device links are managed by the driver core in principle, their creators are allowed and sometimes even expected to drop references to them via device_link_del() or device_link_remove(), but that doesn't really play well with the "persistent" link concept. If "persistent" managed device links are created from driver probe callbacks, device_link_add() called to do that will take a new reference on the link each time the callback runs and those references will never be dropped, which kind of isn't nice. This issues arises because of the link reference counting carried out by device_link_add() for existing links, but that is only done to avoid deleting device links that may still be necessary, which shouldn't be a concern for managed (stateful) links. These device links are managed by the driver core and whoever creates one of them will need it at least as long as until the consumer driver is detached from its device and deleting it may be left to the driver core just fine. For this reason, rework device_link_add() to apply the reference counting to stateless links only and make device_link_del() and device_link_remove() drop references to stateless links only too. After this change, if called to add a stateful device link for a consumer-supplier pair for which a stateful device link is present already, device_link_add() will return the existing link without incrementing its reference counter. Accordingly, device_link_del() and device_link_remove() will WARN() and do nothing when called to drop a reference to a stateful link. Thus, effectively, all stateful device links will be owned by the driver core. In addition, clean up the handling of the link management flags, DL_FLAG_AUTOREMOVE_CONSUMER and DL_FLAG_AUTOREMOVE_SUPPLIER, so that (a) they are never set at the same time and (b) if device_link_add() is called for a consumer-supplier pair with an existing stateful link between them, the flags of that link will be combined with the flags passed to device_link_add() to ensure that the life time of the link is sufficient for all of the callers of device_link_add() for the same consumer-supplier pair. Update the device_link_add() kerneldoc comment to reflect the above changes. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-02-01 00:58:33 +00:00
if (flags & DL_FLAG_STATELESS) {
kref_get(&link->kref);
if (link->flags & DL_FLAG_SYNC_STATE_ONLY &&
!(link->flags & DL_FLAG_STATELESS)) {
link->flags |= DL_FLAG_STATELESS;
goto reorder;
} else {
link->flags |= DL_FLAG_STATELESS;
goto out;
}
driver core: Make driver core own stateful device links Even though stateful device links are managed by the driver core in principle, their creators are allowed and sometimes even expected to drop references to them via device_link_del() or device_link_remove(), but that doesn't really play well with the "persistent" link concept. If "persistent" managed device links are created from driver probe callbacks, device_link_add() called to do that will take a new reference on the link each time the callback runs and those references will never be dropped, which kind of isn't nice. This issues arises because of the link reference counting carried out by device_link_add() for existing links, but that is only done to avoid deleting device links that may still be necessary, which shouldn't be a concern for managed (stateful) links. These device links are managed by the driver core and whoever creates one of them will need it at least as long as until the consumer driver is detached from its device and deleting it may be left to the driver core just fine. For this reason, rework device_link_add() to apply the reference counting to stateless links only and make device_link_del() and device_link_remove() drop references to stateless links only too. After this change, if called to add a stateful device link for a consumer-supplier pair for which a stateful device link is present already, device_link_add() will return the existing link without incrementing its reference counter. Accordingly, device_link_del() and device_link_remove() will WARN() and do nothing when called to drop a reference to a stateful link. Thus, effectively, all stateful device links will be owned by the driver core. In addition, clean up the handling of the link management flags, DL_FLAG_AUTOREMOVE_CONSUMER and DL_FLAG_AUTOREMOVE_SUPPLIER, so that (a) they are never set at the same time and (b) if device_link_add() is called for a consumer-supplier pair with an existing stateful link between them, the flags of that link will be combined with the flags passed to device_link_add() to ensure that the life time of the link is sufficient for all of the callers of device_link_add() for the same consumer-supplier pair. Update the device_link_add() kerneldoc comment to reflect the above changes. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-02-01 00:58:33 +00:00
}
/*
* If the life time of the link following from the new flags is
* longer than indicated by the flags of the existing link,
* update the existing link to stay around longer.
*/
if (flags & DL_FLAG_AUTOREMOVE_SUPPLIER) {
if (link->flags & DL_FLAG_AUTOREMOVE_CONSUMER) {
link->flags &= ~DL_FLAG_AUTOREMOVE_CONSUMER;
link->flags |= DL_FLAG_AUTOREMOVE_SUPPLIER;
}
} else if (!(flags & DL_FLAG_AUTOREMOVE_CONSUMER)) {
link->flags &= ~(DL_FLAG_AUTOREMOVE_CONSUMER |
DL_FLAG_AUTOREMOVE_SUPPLIER);
}
driver core: Remove device link creation limitation If device_link_add() is called for a consumer/supplier pair with an existing device link between them and the existing link's type is not in agreement with the flags passed to that function by its caller, NULL will be returned. That is seriously inconvenient, because it forces the callers of device_link_add() to worry about what others may or may not do even if that is not relevant to them for any other reasons. It turns out, however, that this limitation can be made go away relatively easily. The underlying observation is that if DL_FLAG_STATELESS has been passed to device_link_add() in flags for the given consumer/supplier pair at least once, calling either device_link_del() or device_link_remove() to release the link returned by it should work, but there are no other requirements associated with that flag. In turn, if at least one of the callers of device_link_add() for the given consumer/supplier pair has not passed DL_FLAG_STATELESS to it in flags, the driver core should track the status of the link and act on it as appropriate (ie. the link should be treated as "managed"). This means that DL_FLAG_STATELESS needs to be set for managed device links and it should be valid to call device_link_del() or device_link_remove() to drop references to them in certain sutiations. To allow that to happen, introduce a new (internal) device link flag called DL_FLAG_MANAGED and make device_link_add() set it automatically whenever DL_FLAG_STATELESS is not passed to it. Also make it take additional references to existing device links that were previously stateless (that is, with DL_FLAG_STATELESS set and DL_FLAG_MANAGED unset) and will need to be managed going forward and initialize their status (which has been DL_STATE_NONE so far). Accordingly, when a managed device link is dropped automatically by the driver core, make it clear DL_FLAG_MANAGED, reset the link's status back to DL_STATE_NONE and drop the reference to it associated with DL_FLAG_MANAGED instead of just deleting it right away (to allow it to stay around in case it still needs to be released explicitly by someone). With that, since setting DL_FLAG_STATELESS doesn't mean that the device link in question is not managed any more, replace all of the status-tracking checks against DL_FLAG_STATELESS with analogous checks against DL_FLAG_MANAGED and update the documentation to reflect these changes. While at it, make device_link_add() reject flags that it does not recognize, including DL_FLAG_MANAGED. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Reviewed-by: Saravana Kannan <saravanak@google.com> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Review-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/2305283.AStDPdUUnE@kreacher Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-07-16 15:21:06 +00:00
if (!(link->flags & DL_FLAG_MANAGED)) {
kref_get(&link->kref);
link->flags |= DL_FLAG_MANAGED;
device_link_init_status(link, consumer, supplier);
}
if (link->flags & DL_FLAG_SYNC_STATE_ONLY &&
!(flags & DL_FLAG_SYNC_STATE_ONLY)) {
link->flags &= ~DL_FLAG_SYNC_STATE_ONLY;
goto reorder;
}
goto out;
}
link = kzalloc(sizeof(*link), GFP_KERNEL);
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
if (!link)
goto out;
driver core: Fix handling of runtime PM flags in device_link_add() After commit ead18c23c263 ("driver core: Introduce device links reference counting"), if there is a link between the given supplier and the given consumer already, device_link_add() will refcount it and return it unconditionally without updating its flags. It is possible, however, that the second (or any subsequent) caller of device_link_add() for the same consumer-supplier pair will pass DL_FLAG_PM_RUNTIME, possibly along with DL_FLAG_RPM_ACTIVE, in flags to it and the existing link may not behave as expected then. First, if DL_FLAG_PM_RUNTIME is not set in the existing link's flags at all, it needs to be set like during the original initialization of the link. Second, if DL_FLAG_RPM_ACTIVE is passed to device_link_add() in flags (in addition to DL_FLAG_PM_RUNTIME), the existing link should to be updated to reflect the "active" runtime PM configuration of the consumer-supplier pair and extra care must be taken here to avoid possible destructive races with runtime PM of the consumer. To that end, redefine the rpm_active field in struct device_link as a refcount, initialize it to 1 and make rpm_resume() (for the consumer) and device_link_add() increment it whenever they acquire a runtime PM reference on the supplier device. Accordingly, make rpm_suspend() (for the consumer) and pm_runtime_clean_up_links() decrement it and drop runtime PM references to the supplier device in a loop until rpm_active becones 1 again. Fixes: ead18c23c263 ("driver core: Introduce device links reference counting") Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-02-01 00:49:14 +00:00
refcount_set(&link->rpm_active, 1);
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
get_device(supplier);
link->supplier = supplier;
INIT_LIST_HEAD(&link->s_node);
get_device(consumer);
link->consumer = consumer;
INIT_LIST_HEAD(&link->c_node);
link->flags = flags;
kref_init(&link->kref);
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
link->link_dev.class = &devlink_class;
device_set_pm_not_required(&link->link_dev);
dev_set_name(&link->link_dev, "%s:%s--%s:%s",
dev_bus_name(supplier), dev_name(supplier),
dev_bus_name(consumer), dev_name(consumer));
if (device_register(&link->link_dev)) {
driver core: Fix possible memory leak in device_link_add() I got memory leak as follows: unreferenced object 0xffff88801f0b2200 (size 64): comm "i2c-lis2hh12-21", pid 5455, jiffies 4294944606 (age 15.224s) hex dump (first 32 bytes): 72 65 67 75 6c 61 74 6f 72 3a 72 65 67 75 6c 61 regulator:regula 74 6f 72 2e 30 2d 2d 69 32 63 3a 31 2d 30 30 31 tor.0--i2c:1-001 backtrace: [<00000000bf5b0c3b>] __kmalloc_track_caller+0x19f/0x3a0 [<0000000050da42d9>] kvasprintf+0xb5/0x150 [<000000004bbbed13>] kvasprintf_const+0x60/0x190 [<00000000cdac7480>] kobject_set_name_vargs+0x56/0x150 [<00000000bf83f8e8>] dev_set_name+0xc0/0x100 [<00000000cc1cf7e3>] device_link_add+0x6b4/0x17c0 [<000000009db9faed>] _regulator_get+0x297/0x680 [<00000000845e7f2b>] _devm_regulator_get+0x5b/0xe0 [<000000003958ee25>] st_sensors_power_enable+0x71/0x1b0 [st_sensors] [<000000005f450f52>] st_accel_i2c_probe+0xd9/0x150 [st_accel_i2c] [<00000000b5f2ab33>] i2c_device_probe+0x4d8/0xbe0 [<0000000070fb977b>] really_probe+0x299/0xc30 [<0000000088e226ce>] __driver_probe_device+0x357/0x500 [<00000000c21dda32>] driver_probe_device+0x4e/0x140 [<000000004e650441>] __device_attach_driver+0x257/0x340 [<00000000cf1891b8>] bus_for_each_drv+0x166/0x1e0 When device_register() returns an error, the name allocated in dev_set_name() will be leaked, the put_device() should be used instead of kfree() to give up the device reference, then the name will be freed in kobject_cleanup() and the references of consumer and supplier will be decreased in device_link_release_fn(). Fixes: 287905e68dd2 ("driver core: Expose device link details in sysfs") Reported-by: Hulk Robot <hulkci@huawei.com> Reviewed-by: Saravana Kannan <saravanak@google.com> Reviewed-by: Rafael J. Wysocki <rafael@kernel.org> Signed-off-by: Yang Yingliang <yangyingliang@huawei.com> Link: https://lore.kernel.org/r/20210930085714.2057460-1-yangyingliang@huawei.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2021-09-30 08:57:14 +00:00
put_device(&link->link_dev);
link = NULL;
goto out;
}
if (flags & DL_FLAG_PM_RUNTIME) {
if (flags & DL_FLAG_RPM_ACTIVE)
refcount_inc(&link->rpm_active);
pm_runtime_new_link(consumer);
}
/* Determine the initial link state. */
driver core: Remove device link creation limitation If device_link_add() is called for a consumer/supplier pair with an existing device link between them and the existing link's type is not in agreement with the flags passed to that function by its caller, NULL will be returned. That is seriously inconvenient, because it forces the callers of device_link_add() to worry about what others may or may not do even if that is not relevant to them for any other reasons. It turns out, however, that this limitation can be made go away relatively easily. The underlying observation is that if DL_FLAG_STATELESS has been passed to device_link_add() in flags for the given consumer/supplier pair at least once, calling either device_link_del() or device_link_remove() to release the link returned by it should work, but there are no other requirements associated with that flag. In turn, if at least one of the callers of device_link_add() for the given consumer/supplier pair has not passed DL_FLAG_STATELESS to it in flags, the driver core should track the status of the link and act on it as appropriate (ie. the link should be treated as "managed"). This means that DL_FLAG_STATELESS needs to be set for managed device links and it should be valid to call device_link_del() or device_link_remove() to drop references to them in certain sutiations. To allow that to happen, introduce a new (internal) device link flag called DL_FLAG_MANAGED and make device_link_add() set it automatically whenever DL_FLAG_STATELESS is not passed to it. Also make it take additional references to existing device links that were previously stateless (that is, with DL_FLAG_STATELESS set and DL_FLAG_MANAGED unset) and will need to be managed going forward and initialize their status (which has been DL_STATE_NONE so far). Accordingly, when a managed device link is dropped automatically by the driver core, make it clear DL_FLAG_MANAGED, reset the link's status back to DL_STATE_NONE and drop the reference to it associated with DL_FLAG_MANAGED instead of just deleting it right away (to allow it to stay around in case it still needs to be released explicitly by someone). With that, since setting DL_FLAG_STATELESS doesn't mean that the device link in question is not managed any more, replace all of the status-tracking checks against DL_FLAG_STATELESS with analogous checks against DL_FLAG_MANAGED and update the documentation to reflect these changes. While at it, make device_link_add() reject flags that it does not recognize, including DL_FLAG_MANAGED. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Reviewed-by: Saravana Kannan <saravanak@google.com> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Review-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/2305283.AStDPdUUnE@kreacher Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-07-16 15:21:06 +00:00
if (flags & DL_FLAG_STATELESS)
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
link->status = DL_STATE_NONE;
driver core: Remove device link creation limitation If device_link_add() is called for a consumer/supplier pair with an existing device link between them and the existing link's type is not in agreement with the flags passed to that function by its caller, NULL will be returned. That is seriously inconvenient, because it forces the callers of device_link_add() to worry about what others may or may not do even if that is not relevant to them for any other reasons. It turns out, however, that this limitation can be made go away relatively easily. The underlying observation is that if DL_FLAG_STATELESS has been passed to device_link_add() in flags for the given consumer/supplier pair at least once, calling either device_link_del() or device_link_remove() to release the link returned by it should work, but there are no other requirements associated with that flag. In turn, if at least one of the callers of device_link_add() for the given consumer/supplier pair has not passed DL_FLAG_STATELESS to it in flags, the driver core should track the status of the link and act on it as appropriate (ie. the link should be treated as "managed"). This means that DL_FLAG_STATELESS needs to be set for managed device links and it should be valid to call device_link_del() or device_link_remove() to drop references to them in certain sutiations. To allow that to happen, introduce a new (internal) device link flag called DL_FLAG_MANAGED and make device_link_add() set it automatically whenever DL_FLAG_STATELESS is not passed to it. Also make it take additional references to existing device links that were previously stateless (that is, with DL_FLAG_STATELESS set and DL_FLAG_MANAGED unset) and will need to be managed going forward and initialize their status (which has been DL_STATE_NONE so far). Accordingly, when a managed device link is dropped automatically by the driver core, make it clear DL_FLAG_MANAGED, reset the link's status back to DL_STATE_NONE and drop the reference to it associated with DL_FLAG_MANAGED instead of just deleting it right away (to allow it to stay around in case it still needs to be released explicitly by someone). With that, since setting DL_FLAG_STATELESS doesn't mean that the device link in question is not managed any more, replace all of the status-tracking checks against DL_FLAG_STATELESS with analogous checks against DL_FLAG_MANAGED and update the documentation to reflect these changes. While at it, make device_link_add() reject flags that it does not recognize, including DL_FLAG_MANAGED. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Reviewed-by: Saravana Kannan <saravanak@google.com> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Review-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/2305283.AStDPdUUnE@kreacher Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-07-16 15:21:06 +00:00
else
device_link_init_status(link, consumer, supplier);
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
driver core: Fix adding device links to probing suppliers Currently, it is not valid to add a device link from a consumer driver ->probe callback to a supplier that is still probing too, but generally this is a valid use case. For example, if the consumer has just acquired a resource that can only be available if the supplier is functional, adding a device link to that supplier right away should be safe (and even desirable arguably), but device_link_add() doesn't handle that case correctly and the initial state of the link created by it is wrong then. To address this problem, change the initial state of device links added between a probing supplier and a probing consumer to DL_STATE_CONSUMER_PROBE and update device_links_driver_bound() to skip such links on the supplier side. With this change, if the supplier probe completes first, device_links_driver_bound() called for it will skip the link state update and when it is called for the consumer, the link state will be updated to "active". In turn, if the consumer probe completes first, device_links_driver_bound() called for it will change the state of the link to "active" and when it is called for the supplier, the link status update will be skipped. However, in principle the supplier or consumer probe may still fail after the link has been added, so modify device_links_no_driver() to change device links in the "active" or "consumer probe" state to "dormant" on the supplier side and update __device_links_no_driver() to change the link state to "available" only if it is "consumer probe" or "active". Then, if the supplier probe fails first, the leftover link to the probing consumer will become "dormant" and device_links_no_driver() called for the consumer (when its probe fails) will clean it up. In turn, if the consumer probe fails first, it will either drop the link, or change its state to "available" and, in the latter case, when device_links_no_driver() is called for the supplier, it will update the link state to "dormant". [If the supplier probe fails, but the consumer probe succeeds, which should not happen as long as the consumer driver is correct, the link still will be around, but it will be "dormant" until the supplier is probed again.] Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-02-01 00:50:39 +00:00
/*
* Some callers expect the link creation during consumer driver probe to
* resume the supplier even without DL_FLAG_RPM_ACTIVE.
*/
if (link->status == DL_STATE_CONSUMER_PROBE &&
flags & DL_FLAG_PM_RUNTIME)
pm_runtime_resume(supplier);
driver core: Fix SYNC_STATE_ONLY device link implementation When SYNC_STATE_ONLY support was added in commit 05ef983e0d65 ("driver core: Add device link support for SYNC_STATE_ONLY flag"), device_link_add() incorrectly skipped adding the new SYNC_STATE_ONLY device link to the supplier's and consumer's "device link" list. This causes multiple issues: - The device link is lost forever from driver core if the caller didn't keep track of it (caller typically isn't expected to). This is a memory leak. - The device link is also never visible to any other code path after device_link_add() returns. If we fix the "device link" list handling, that exposes a bunch of issues. 1. The device link "status" state management code rightfully doesn't handle the case where a DL_FLAG_MANAGED device link exists between a supplier and consumer, but the consumer manages to probe successfully before the supplier. The addition of DL_FLAG_SYNC_STATE_ONLY links break this assumption. This causes device_links_driver_bound() to throw a warning when this happens. Since DL_FLAG_SYNC_STATE_ONLY device links are mainly used for creating proxy device links for child device dependencies and aren't useful once the consumer device probes successfully, this patch just deletes DL_FLAG_SYNC_STATE_ONLY device links once its consumer device probes. This way, we avoid the warning, free up some memory and avoid complicating the device links "status" state management code. 2. Creating a DL_FLAG_STATELESS device link between two devices that already have a DL_FLAG_SYNC_STATE_ONLY device link will result in the DL_FLAG_STATELESS flag not getting set correctly. This patch also fixes this. Lastly, this patch also fixes minor whitespace issues. Cc: stable@vger.kernel.org Fixes: 05ef983e0d65 ("driver core: Add device link support for SYNC_STATE_ONLY flag") Signed-off-by: Saravana Kannan <saravanak@google.com> Reviewed-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Link: https://lore.kernel.org/r/20200519063000.128819-1-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-05-19 06:30:00 +00:00
list_add_tail_rcu(&link->s_node, &supplier->links.consumers);
list_add_tail_rcu(&link->c_node, &consumer->links.suppliers);
if (flags & DL_FLAG_SYNC_STATE_ONLY) {
dev_dbg(consumer,
"Linked as a sync state only consumer to %s\n",
dev_name(supplier));
goto out;
}
driver core: Fix SYNC_STATE_ONLY device link implementation When SYNC_STATE_ONLY support was added in commit 05ef983e0d65 ("driver core: Add device link support for SYNC_STATE_ONLY flag"), device_link_add() incorrectly skipped adding the new SYNC_STATE_ONLY device link to the supplier's and consumer's "device link" list. This causes multiple issues: - The device link is lost forever from driver core if the caller didn't keep track of it (caller typically isn't expected to). This is a memory leak. - The device link is also never visible to any other code path after device_link_add() returns. If we fix the "device link" list handling, that exposes a bunch of issues. 1. The device link "status" state management code rightfully doesn't handle the case where a DL_FLAG_MANAGED device link exists between a supplier and consumer, but the consumer manages to probe successfully before the supplier. The addition of DL_FLAG_SYNC_STATE_ONLY links break this assumption. This causes device_links_driver_bound() to throw a warning when this happens. Since DL_FLAG_SYNC_STATE_ONLY device links are mainly used for creating proxy device links for child device dependencies and aren't useful once the consumer device probes successfully, this patch just deletes DL_FLAG_SYNC_STATE_ONLY device links once its consumer device probes. This way, we avoid the warning, free up some memory and avoid complicating the device links "status" state management code. 2. Creating a DL_FLAG_STATELESS device link between two devices that already have a DL_FLAG_SYNC_STATE_ONLY device link will result in the DL_FLAG_STATELESS flag not getting set correctly. This patch also fixes this. Lastly, this patch also fixes minor whitespace issues. Cc: stable@vger.kernel.org Fixes: 05ef983e0d65 ("driver core: Add device link support for SYNC_STATE_ONLY flag") Signed-off-by: Saravana Kannan <saravanak@google.com> Reviewed-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Link: https://lore.kernel.org/r/20200519063000.128819-1-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-05-19 06:30:00 +00:00
reorder:
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
/*
* Move the consumer and all of the devices depending on it to the end
* of dpm_list and the devices_kset list.
*
* It is necessary to hold dpm_list locked throughout all that or else
* we may end up suspending with a wrong ordering of it.
*/
device_reorder_to_tail(consumer, NULL);
dev_dbg(consumer, "Linked as a consumer to %s\n", dev_name(supplier));
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
driver core: Fix SYNC_STATE_ONLY device link implementation When SYNC_STATE_ONLY support was added in commit 05ef983e0d65 ("driver core: Add device link support for SYNC_STATE_ONLY flag"), device_link_add() incorrectly skipped adding the new SYNC_STATE_ONLY device link to the supplier's and consumer's "device link" list. This causes multiple issues: - The device link is lost forever from driver core if the caller didn't keep track of it (caller typically isn't expected to). This is a memory leak. - The device link is also never visible to any other code path after device_link_add() returns. If we fix the "device link" list handling, that exposes a bunch of issues. 1. The device link "status" state management code rightfully doesn't handle the case where a DL_FLAG_MANAGED device link exists between a supplier and consumer, but the consumer manages to probe successfully before the supplier. The addition of DL_FLAG_SYNC_STATE_ONLY links break this assumption. This causes device_links_driver_bound() to throw a warning when this happens. Since DL_FLAG_SYNC_STATE_ONLY device links are mainly used for creating proxy device links for child device dependencies and aren't useful once the consumer device probes successfully, this patch just deletes DL_FLAG_SYNC_STATE_ONLY device links once its consumer device probes. This way, we avoid the warning, free up some memory and avoid complicating the device links "status" state management code. 2. Creating a DL_FLAG_STATELESS device link between two devices that already have a DL_FLAG_SYNC_STATE_ONLY device link will result in the DL_FLAG_STATELESS flag not getting set correctly. This patch also fixes this. Lastly, this patch also fixes minor whitespace issues. Cc: stable@vger.kernel.org Fixes: 05ef983e0d65 ("driver core: Add device link support for SYNC_STATE_ONLY flag") Signed-off-by: Saravana Kannan <saravanak@google.com> Reviewed-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Link: https://lore.kernel.org/r/20200519063000.128819-1-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-05-19 06:30:00 +00:00
out:
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
device_pm_unlock();
device_links_write_unlock();
driver core: Fix handling of runtime PM flags in device_link_add() After commit ead18c23c263 ("driver core: Introduce device links reference counting"), if there is a link between the given supplier and the given consumer already, device_link_add() will refcount it and return it unconditionally without updating its flags. It is possible, however, that the second (or any subsequent) caller of device_link_add() for the same consumer-supplier pair will pass DL_FLAG_PM_RUNTIME, possibly along with DL_FLAG_RPM_ACTIVE, in flags to it and the existing link may not behave as expected then. First, if DL_FLAG_PM_RUNTIME is not set in the existing link's flags at all, it needs to be set like during the original initialization of the link. Second, if DL_FLAG_RPM_ACTIVE is passed to device_link_add() in flags (in addition to DL_FLAG_PM_RUNTIME), the existing link should to be updated to reflect the "active" runtime PM configuration of the consumer-supplier pair and extra care must be taken here to avoid possible destructive races with runtime PM of the consumer. To that end, redefine the rpm_active field in struct device_link as a refcount, initialize it to 1 and make rpm_resume() (for the consumer) and device_link_add() increment it whenever they acquire a runtime PM reference on the supplier device. Accordingly, make rpm_suspend() (for the consumer) and pm_runtime_clean_up_links() decrement it and drop runtime PM references to the supplier device in a loop until rpm_active becones 1 again. Fixes: ead18c23c263 ("driver core: Introduce device links reference counting") Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-02-01 00:49:14 +00:00
if ((flags & DL_FLAG_PM_RUNTIME && flags & DL_FLAG_RPM_ACTIVE) && !link)
pm_runtime_put(supplier);
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
return link;
}
EXPORT_SYMBOL_GPL(device_link_add);
static void __device_link_del(struct kref *kref)
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
{
struct device_link *link = container_of(kref, struct device_link, kref);
dev_dbg(link->consumer, "Dropping the link to %s\n",
dev_name(link->supplier));
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
pm_runtime_drop_link(link);
device_link_remove_from_lists(link);
driver core: Fix sleeping in invalid context during device link deletion Marek and Guenter reported that commit 287905e68dd2 ("driver core: Expose device link details in sysfs") caused sleeping/scheduling while atomic warnings. BUG: sleeping function called from invalid context at kernel/locking/mutex.c:935 in_atomic(): 1, irqs_disabled(): 0, non_block: 0, pid: 12, name: kworker/0:1 2 locks held by kworker/0:1/12: #0: ee8074a8 ((wq_completion)rcu_gp){+.+.}-{0:0}, at: process_one_work+0x174/0x7dc #1: ee921f20 ((work_completion)(&sdp->work)){+.+.}-{0:0}, at: process_one_work+0x174/0x7dc Preemption disabled at: [<c01b10f0>] srcu_invoke_callbacks+0xc0/0x154 ----- 8< ----- SNIP [<c064590c>] (device_del) from [<c0645c9c>] (device_unregister+0x24/0x64) [<c0645c9c>] (device_unregister) from [<c01b10fc>] (srcu_invoke_callbacks+0xcc/0x154) [<c01b10fc>] (srcu_invoke_callbacks) from [<c01493c4>] (process_one_work+0x234/0x7dc) [<c01493c4>] (process_one_work) from [<c01499b0>] (worker_thread+0x44/0x51c) [<c01499b0>] (worker_thread) from [<c0150bf4>] (kthread+0x158/0x1a0) [<c0150bf4>] (kthread) from [<c0100114>] (ret_from_fork+0x14/0x20) Exception stack(0xee921fb0 to 0xee921ff8) This was caused by the device link device being released in the context of srcu_invoke_callbacks(). There is no need to wait till the RCU callback to release the device link device. So release the device earlier and move the call_srcu() into the device release code. That way, the memory will get freed only after the device is released AND the RCU callback is called. Fixes: 287905e68dd2 ("driver core: Expose device link details in sysfs") Reported-by: Marek Szyprowski <m.szyprowski@samsung.com> Reported-by: Guenter Roeck <linux@roeck-us.net> Reported-by: Naresh Kamboju <naresh.kamboju@linaro.org> Signed-off-by: Saravana Kannan <saravanak@google.com> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Tested-by: Guenter Roeck <linux@roeck-us.net> Link: https://lore.kernel.org/r/20200716214523.2924704-1-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-07-16 21:45:23 +00:00
device_unregister(&link->link_dev);
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
}
driver core: Make driver core own stateful device links Even though stateful device links are managed by the driver core in principle, their creators are allowed and sometimes even expected to drop references to them via device_link_del() or device_link_remove(), but that doesn't really play well with the "persistent" link concept. If "persistent" managed device links are created from driver probe callbacks, device_link_add() called to do that will take a new reference on the link each time the callback runs and those references will never be dropped, which kind of isn't nice. This issues arises because of the link reference counting carried out by device_link_add() for existing links, but that is only done to avoid deleting device links that may still be necessary, which shouldn't be a concern for managed (stateful) links. These device links are managed by the driver core and whoever creates one of them will need it at least as long as until the consumer driver is detached from its device and deleting it may be left to the driver core just fine. For this reason, rework device_link_add() to apply the reference counting to stateless links only and make device_link_del() and device_link_remove() drop references to stateless links only too. After this change, if called to add a stateful device link for a consumer-supplier pair for which a stateful device link is present already, device_link_add() will return the existing link without incrementing its reference counter. Accordingly, device_link_del() and device_link_remove() will WARN() and do nothing when called to drop a reference to a stateful link. Thus, effectively, all stateful device links will be owned by the driver core. In addition, clean up the handling of the link management flags, DL_FLAG_AUTOREMOVE_CONSUMER and DL_FLAG_AUTOREMOVE_SUPPLIER, so that (a) they are never set at the same time and (b) if device_link_add() is called for a consumer-supplier pair with an existing stateful link between them, the flags of that link will be combined with the flags passed to device_link_add() to ensure that the life time of the link is sufficient for all of the callers of device_link_add() for the same consumer-supplier pair. Update the device_link_add() kerneldoc comment to reflect the above changes. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-02-01 00:58:33 +00:00
static void device_link_put_kref(struct device_link *link)
{
if (link->flags & DL_FLAG_STATELESS)
kref_put(&link->kref, __device_link_del);
else if (!device_is_registered(link->consumer))
__device_link_del(&link->kref);
driver core: Make driver core own stateful device links Even though stateful device links are managed by the driver core in principle, their creators are allowed and sometimes even expected to drop references to them via device_link_del() or device_link_remove(), but that doesn't really play well with the "persistent" link concept. If "persistent" managed device links are created from driver probe callbacks, device_link_add() called to do that will take a new reference on the link each time the callback runs and those references will never be dropped, which kind of isn't nice. This issues arises because of the link reference counting carried out by device_link_add() for existing links, but that is only done to avoid deleting device links that may still be necessary, which shouldn't be a concern for managed (stateful) links. These device links are managed by the driver core and whoever creates one of them will need it at least as long as until the consumer driver is detached from its device and deleting it may be left to the driver core just fine. For this reason, rework device_link_add() to apply the reference counting to stateless links only and make device_link_del() and device_link_remove() drop references to stateless links only too. After this change, if called to add a stateful device link for a consumer-supplier pair for which a stateful device link is present already, device_link_add() will return the existing link without incrementing its reference counter. Accordingly, device_link_del() and device_link_remove() will WARN() and do nothing when called to drop a reference to a stateful link. Thus, effectively, all stateful device links will be owned by the driver core. In addition, clean up the handling of the link management flags, DL_FLAG_AUTOREMOVE_CONSUMER and DL_FLAG_AUTOREMOVE_SUPPLIER, so that (a) they are never set at the same time and (b) if device_link_add() is called for a consumer-supplier pair with an existing stateful link between them, the flags of that link will be combined with the flags passed to device_link_add() to ensure that the life time of the link is sufficient for all of the callers of device_link_add() for the same consumer-supplier pair. Update the device_link_add() kerneldoc comment to reflect the above changes. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-02-01 00:58:33 +00:00
else
WARN(1, "Unable to drop a managed device link reference\n");
}
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
/**
driver core: Make driver core own stateful device links Even though stateful device links are managed by the driver core in principle, their creators are allowed and sometimes even expected to drop references to them via device_link_del() or device_link_remove(), but that doesn't really play well with the "persistent" link concept. If "persistent" managed device links are created from driver probe callbacks, device_link_add() called to do that will take a new reference on the link each time the callback runs and those references will never be dropped, which kind of isn't nice. This issues arises because of the link reference counting carried out by device_link_add() for existing links, but that is only done to avoid deleting device links that may still be necessary, which shouldn't be a concern for managed (stateful) links. These device links are managed by the driver core and whoever creates one of them will need it at least as long as until the consumer driver is detached from its device and deleting it may be left to the driver core just fine. For this reason, rework device_link_add() to apply the reference counting to stateless links only and make device_link_del() and device_link_remove() drop references to stateless links only too. After this change, if called to add a stateful device link for a consumer-supplier pair for which a stateful device link is present already, device_link_add() will return the existing link without incrementing its reference counter. Accordingly, device_link_del() and device_link_remove() will WARN() and do nothing when called to drop a reference to a stateful link. Thus, effectively, all stateful device links will be owned by the driver core. In addition, clean up the handling of the link management flags, DL_FLAG_AUTOREMOVE_CONSUMER and DL_FLAG_AUTOREMOVE_SUPPLIER, so that (a) they are never set at the same time and (b) if device_link_add() is called for a consumer-supplier pair with an existing stateful link between them, the flags of that link will be combined with the flags passed to device_link_add() to ensure that the life time of the link is sufficient for all of the callers of device_link_add() for the same consumer-supplier pair. Update the device_link_add() kerneldoc comment to reflect the above changes. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-02-01 00:58:33 +00:00
* device_link_del - Delete a stateless link between two devices.
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
* @link: Device link to delete.
*
* The caller must ensure proper synchronization of this function with runtime
* PM. If the link was added multiple times, it needs to be deleted as often.
* Care is required for hotplugged devices: Their links are purged on removal
* and calling device_link_del() is then no longer allowed.
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
*/
void device_link_del(struct device_link *link)
{
device_links_write_lock();
driver core: Make driver core own stateful device links Even though stateful device links are managed by the driver core in principle, their creators are allowed and sometimes even expected to drop references to them via device_link_del() or device_link_remove(), but that doesn't really play well with the "persistent" link concept. If "persistent" managed device links are created from driver probe callbacks, device_link_add() called to do that will take a new reference on the link each time the callback runs and those references will never be dropped, which kind of isn't nice. This issues arises because of the link reference counting carried out by device_link_add() for existing links, but that is only done to avoid deleting device links that may still be necessary, which shouldn't be a concern for managed (stateful) links. These device links are managed by the driver core and whoever creates one of them will need it at least as long as until the consumer driver is detached from its device and deleting it may be left to the driver core just fine. For this reason, rework device_link_add() to apply the reference counting to stateless links only and make device_link_del() and device_link_remove() drop references to stateless links only too. After this change, if called to add a stateful device link for a consumer-supplier pair for which a stateful device link is present already, device_link_add() will return the existing link without incrementing its reference counter. Accordingly, device_link_del() and device_link_remove() will WARN() and do nothing when called to drop a reference to a stateful link. Thus, effectively, all stateful device links will be owned by the driver core. In addition, clean up the handling of the link management flags, DL_FLAG_AUTOREMOVE_CONSUMER and DL_FLAG_AUTOREMOVE_SUPPLIER, so that (a) they are never set at the same time and (b) if device_link_add() is called for a consumer-supplier pair with an existing stateful link between them, the flags of that link will be combined with the flags passed to device_link_add() to ensure that the life time of the link is sufficient for all of the callers of device_link_add() for the same consumer-supplier pair. Update the device_link_add() kerneldoc comment to reflect the above changes. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-02-01 00:58:33 +00:00
device_link_put_kref(link);
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
device_links_write_unlock();
}
EXPORT_SYMBOL_GPL(device_link_del);
/**
driver core: Make driver core own stateful device links Even though stateful device links are managed by the driver core in principle, their creators are allowed and sometimes even expected to drop references to them via device_link_del() or device_link_remove(), but that doesn't really play well with the "persistent" link concept. If "persistent" managed device links are created from driver probe callbacks, device_link_add() called to do that will take a new reference on the link each time the callback runs and those references will never be dropped, which kind of isn't nice. This issues arises because of the link reference counting carried out by device_link_add() for existing links, but that is only done to avoid deleting device links that may still be necessary, which shouldn't be a concern for managed (stateful) links. These device links are managed by the driver core and whoever creates one of them will need it at least as long as until the consumer driver is detached from its device and deleting it may be left to the driver core just fine. For this reason, rework device_link_add() to apply the reference counting to stateless links only and make device_link_del() and device_link_remove() drop references to stateless links only too. After this change, if called to add a stateful device link for a consumer-supplier pair for which a stateful device link is present already, device_link_add() will return the existing link without incrementing its reference counter. Accordingly, device_link_del() and device_link_remove() will WARN() and do nothing when called to drop a reference to a stateful link. Thus, effectively, all stateful device links will be owned by the driver core. In addition, clean up the handling of the link management flags, DL_FLAG_AUTOREMOVE_CONSUMER and DL_FLAG_AUTOREMOVE_SUPPLIER, so that (a) they are never set at the same time and (b) if device_link_add() is called for a consumer-supplier pair with an existing stateful link between them, the flags of that link will be combined with the flags passed to device_link_add() to ensure that the life time of the link is sufficient for all of the callers of device_link_add() for the same consumer-supplier pair. Update the device_link_add() kerneldoc comment to reflect the above changes. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-02-01 00:58:33 +00:00
* device_link_remove - Delete a stateless link between two devices.
* @consumer: Consumer end of the link.
* @supplier: Supplier end of the link.
*
* The caller must ensure proper synchronization of this function with runtime
* PM.
*/
void device_link_remove(void *consumer, struct device *supplier)
{
struct device_link *link;
if (WARN_ON(consumer == supplier))
return;
device_links_write_lock();
list_for_each_entry(link, &supplier->links.consumers, s_node) {
if (link->consumer == consumer) {
driver core: Make driver core own stateful device links Even though stateful device links are managed by the driver core in principle, their creators are allowed and sometimes even expected to drop references to them via device_link_del() or device_link_remove(), but that doesn't really play well with the "persistent" link concept. If "persistent" managed device links are created from driver probe callbacks, device_link_add() called to do that will take a new reference on the link each time the callback runs and those references will never be dropped, which kind of isn't nice. This issues arises because of the link reference counting carried out by device_link_add() for existing links, but that is only done to avoid deleting device links that may still be necessary, which shouldn't be a concern for managed (stateful) links. These device links are managed by the driver core and whoever creates one of them will need it at least as long as until the consumer driver is detached from its device and deleting it may be left to the driver core just fine. For this reason, rework device_link_add() to apply the reference counting to stateless links only and make device_link_del() and device_link_remove() drop references to stateless links only too. After this change, if called to add a stateful device link for a consumer-supplier pair for which a stateful device link is present already, device_link_add() will return the existing link without incrementing its reference counter. Accordingly, device_link_del() and device_link_remove() will WARN() and do nothing when called to drop a reference to a stateful link. Thus, effectively, all stateful device links will be owned by the driver core. In addition, clean up the handling of the link management flags, DL_FLAG_AUTOREMOVE_CONSUMER and DL_FLAG_AUTOREMOVE_SUPPLIER, so that (a) they are never set at the same time and (b) if device_link_add() is called for a consumer-supplier pair with an existing stateful link between them, the flags of that link will be combined with the flags passed to device_link_add() to ensure that the life time of the link is sufficient for all of the callers of device_link_add() for the same consumer-supplier pair. Update the device_link_add() kerneldoc comment to reflect the above changes. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-02-01 00:58:33 +00:00
device_link_put_kref(link);
break;
}
}
device_links_write_unlock();
}
EXPORT_SYMBOL_GPL(device_link_remove);
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
static void device_links_missing_supplier(struct device *dev)
{
struct device_link *link;
driver core: Update device link status correctly for SYNC_STATE_ONLY links When SYNC_STATE_ONLY support was added in commit 05ef983e0d65 ("driver core: Add device link support for SYNC_STATE_ONLY flag"), SYNC_STATE_ONLY links were treated similar to STATELESS links in terms of not blocking consumer probe if the supplier hasn't probed yet. That caused a SYNC_STATE_ONLY device link's status to not get updated. Since SYNC_STATE_ONLY device link is no longer useful once the consumer probes, commit 21c27f06587d ("driver core: Fix SYNC_STATE_ONLY device link implementation") addresses the status update issue by deleting the SYNC_STATE_ONLY device link instead of complicating the status update code. However, there are still some cases where we need to update the status of a SYNC_STATE_ONLY device link. This is because a SYNC_STATE_ONLY device link can later get converted into a normal MANAGED device link when a normal MANAGED device link is created between a supplier and consumer that already have a SYNC_STATE_ONLY device link between them. If a SYNC_STATE_ONLY device link's status isn't maintained correctly till it's converted to a normal MANAGED device link, then the normal MANAGED device link will end up with a wrong link status. This can cause a warning stack trace[1] when the consumer device probes successfully. This commit fixes the SYNC_STATE_ONLY device link status update issue where it wouldn't transition correctly from DL_STATE_DORMANT or DL_STATE_AVAILABLE to DL_STATE_CONSUMER_PROBE. It also resets the status back to DL_STATE_DORMANT or DL_STATE_AVAILABLE if the consumer probe fails. [1] - https://lore.kernel.org/lkml/20200522204120.3b3c9ed6@apollo/ Fixes: 05ef983e0d65 ("driver core: Add device link support for SYNC_STATE_ONLY flag") Fixes: 21c27f06587d ("driver core: Fix SYNC_STATE_ONLY device link implementation") Reported-by: Michael Walle <michael@walle.cc> Tested-by: Michael Walle <michael@walle.cc> Signed-off-by: Saravana Kannan <saravanak@google.com> Reviewed-by: Rafael J. Wysocki <rrafael.j.wysocki@intel.com> Link: https://lore.kernel.org/r/20200526220928.49939-1-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-05-26 22:09:27 +00:00
list_for_each_entry(link, &dev->links.suppliers, c_node) {
if (link->status != DL_STATE_CONSUMER_PROBE)
continue;
if (link->supplier->links.status == DL_DEV_DRIVER_BOUND) {
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
WRITE_ONCE(link->status, DL_STATE_AVAILABLE);
driver core: Update device link status correctly for SYNC_STATE_ONLY links When SYNC_STATE_ONLY support was added in commit 05ef983e0d65 ("driver core: Add device link support for SYNC_STATE_ONLY flag"), SYNC_STATE_ONLY links were treated similar to STATELESS links in terms of not blocking consumer probe if the supplier hasn't probed yet. That caused a SYNC_STATE_ONLY device link's status to not get updated. Since SYNC_STATE_ONLY device link is no longer useful once the consumer probes, commit 21c27f06587d ("driver core: Fix SYNC_STATE_ONLY device link implementation") addresses the status update issue by deleting the SYNC_STATE_ONLY device link instead of complicating the status update code. However, there are still some cases where we need to update the status of a SYNC_STATE_ONLY device link. This is because a SYNC_STATE_ONLY device link can later get converted into a normal MANAGED device link when a normal MANAGED device link is created between a supplier and consumer that already have a SYNC_STATE_ONLY device link between them. If a SYNC_STATE_ONLY device link's status isn't maintained correctly till it's converted to a normal MANAGED device link, then the normal MANAGED device link will end up with a wrong link status. This can cause a warning stack trace[1] when the consumer device probes successfully. This commit fixes the SYNC_STATE_ONLY device link status update issue where it wouldn't transition correctly from DL_STATE_DORMANT or DL_STATE_AVAILABLE to DL_STATE_CONSUMER_PROBE. It also resets the status back to DL_STATE_DORMANT or DL_STATE_AVAILABLE if the consumer probe fails. [1] - https://lore.kernel.org/lkml/20200522204120.3b3c9ed6@apollo/ Fixes: 05ef983e0d65 ("driver core: Add device link support for SYNC_STATE_ONLY flag") Fixes: 21c27f06587d ("driver core: Fix SYNC_STATE_ONLY device link implementation") Reported-by: Michael Walle <michael@walle.cc> Tested-by: Michael Walle <michael@walle.cc> Signed-off-by: Saravana Kannan <saravanak@google.com> Reviewed-by: Rafael J. Wysocki <rrafael.j.wysocki@intel.com> Link: https://lore.kernel.org/r/20200526220928.49939-1-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-05-26 22:09:27 +00:00
} else {
WARN_ON(!(link->flags & DL_FLAG_SYNC_STATE_ONLY));
WRITE_ONCE(link->status, DL_STATE_DORMANT);
}
}
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
}
driver core: Add wait_for_init_devices_probe helper function Some devices might need to be probed and bound successfully before the kernel boot sequence can finish and move on to init/userspace. For example, a network interface might need to be bound to be able to mount a NFS rootfs. With fw_devlink=on by default, some of these devices might be blocked from probing because they are waiting on a optional supplier that doesn't have a driver. While fw_devlink will eventually identify such devices and unblock the probing automatically, it might be too late by the time it unblocks the probing of devices. For example, the IP4 autoconfig might timeout before fw_devlink unblocks probing of the network interface. This function is available to temporarily try and probe all devices that have a driver even if some of their suppliers haven't been added or don't have drivers. The drivers can then decide which of the suppliers are optional vs mandatory and probe the device if possible. By the time this function returns, all such "best effort" probes are guaranteed to be completed. If a device successfully probes in this mode, we delete all fw_devlink discovered dependencies of that device where the supplier hasn't yet probed successfully because they have to be optional dependencies. This also means that some devices that aren't needed for init and could have waited for their optional supplier to probe (when the supplier's module is loaded later on) would end up probing prematurely with limited functionality. So call this function only when boot would fail without it. Tested-by: Geert Uytterhoeven <geert+renesas@glider.be> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20220601070707.3946847-5-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-06-01 07:07:00 +00:00
static bool dev_is_best_effort(struct device *dev)
{
return (fw_devlink_best_effort && dev->can_match) ||
(dev->fwnode && (dev->fwnode->flags & FWNODE_FLAG_BEST_EFFORT));
driver core: Add wait_for_init_devices_probe helper function Some devices might need to be probed and bound successfully before the kernel boot sequence can finish and move on to init/userspace. For example, a network interface might need to be bound to be able to mount a NFS rootfs. With fw_devlink=on by default, some of these devices might be blocked from probing because they are waiting on a optional supplier that doesn't have a driver. While fw_devlink will eventually identify such devices and unblock the probing automatically, it might be too late by the time it unblocks the probing of devices. For example, the IP4 autoconfig might timeout before fw_devlink unblocks probing of the network interface. This function is available to temporarily try and probe all devices that have a driver even if some of their suppliers haven't been added or don't have drivers. The drivers can then decide which of the suppliers are optional vs mandatory and probe the device if possible. By the time this function returns, all such "best effort" probes are guaranteed to be completed. If a device successfully probes in this mode, we delete all fw_devlink discovered dependencies of that device where the supplier hasn't yet probed successfully because they have to be optional dependencies. This also means that some devices that aren't needed for init and could have waited for their optional supplier to probe (when the supplier's module is loaded later on) would end up probing prematurely with limited functionality. So call this function only when boot would fail without it. Tested-by: Geert Uytterhoeven <geert+renesas@glider.be> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20220601070707.3946847-5-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-06-01 07:07:00 +00:00
}
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
/**
* device_links_check_suppliers - Check presence of supplier drivers.
* @dev: Consumer device.
*
* Check links from this device to any suppliers. Walk the list of the device's
* links to suppliers and see if all of them are available. If not, simply
* return -EPROBE_DEFER.
*
* We need to guarantee that the supplier will not go away after the check has
* been positive here. It only can go away in __device_release_driver() and
* that function checks the device's links to consumers. This means we need to
* mark the link as "consumer probe in progress" to make the supplier removal
* wait for us to complete (or bad things may happen).
*
driver core: Remove device link creation limitation If device_link_add() is called for a consumer/supplier pair with an existing device link between them and the existing link's type is not in agreement with the flags passed to that function by its caller, NULL will be returned. That is seriously inconvenient, because it forces the callers of device_link_add() to worry about what others may or may not do even if that is not relevant to them for any other reasons. It turns out, however, that this limitation can be made go away relatively easily. The underlying observation is that if DL_FLAG_STATELESS has been passed to device_link_add() in flags for the given consumer/supplier pair at least once, calling either device_link_del() or device_link_remove() to release the link returned by it should work, but there are no other requirements associated with that flag. In turn, if at least one of the callers of device_link_add() for the given consumer/supplier pair has not passed DL_FLAG_STATELESS to it in flags, the driver core should track the status of the link and act on it as appropriate (ie. the link should be treated as "managed"). This means that DL_FLAG_STATELESS needs to be set for managed device links and it should be valid to call device_link_del() or device_link_remove() to drop references to them in certain sutiations. To allow that to happen, introduce a new (internal) device link flag called DL_FLAG_MANAGED and make device_link_add() set it automatically whenever DL_FLAG_STATELESS is not passed to it. Also make it take additional references to existing device links that were previously stateless (that is, with DL_FLAG_STATELESS set and DL_FLAG_MANAGED unset) and will need to be managed going forward and initialize their status (which has been DL_STATE_NONE so far). Accordingly, when a managed device link is dropped automatically by the driver core, make it clear DL_FLAG_MANAGED, reset the link's status back to DL_STATE_NONE and drop the reference to it associated with DL_FLAG_MANAGED instead of just deleting it right away (to allow it to stay around in case it still needs to be released explicitly by someone). With that, since setting DL_FLAG_STATELESS doesn't mean that the device link in question is not managed any more, replace all of the status-tracking checks against DL_FLAG_STATELESS with analogous checks against DL_FLAG_MANAGED and update the documentation to reflect these changes. While at it, make device_link_add() reject flags that it does not recognize, including DL_FLAG_MANAGED. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Reviewed-by: Saravana Kannan <saravanak@google.com> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Review-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/2305283.AStDPdUUnE@kreacher Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-07-16 15:21:06 +00:00
* Links without the DL_FLAG_MANAGED flag set are ignored.
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
*/
int device_links_check_suppliers(struct device *dev)
{
struct device_link *link;
driver core: Add wait_for_init_devices_probe helper function Some devices might need to be probed and bound successfully before the kernel boot sequence can finish and move on to init/userspace. For example, a network interface might need to be bound to be able to mount a NFS rootfs. With fw_devlink=on by default, some of these devices might be blocked from probing because they are waiting on a optional supplier that doesn't have a driver. While fw_devlink will eventually identify such devices and unblock the probing automatically, it might be too late by the time it unblocks the probing of devices. For example, the IP4 autoconfig might timeout before fw_devlink unblocks probing of the network interface. This function is available to temporarily try and probe all devices that have a driver even if some of their suppliers haven't been added or don't have drivers. The drivers can then decide which of the suppliers are optional vs mandatory and probe the device if possible. By the time this function returns, all such "best effort" probes are guaranteed to be completed. If a device successfully probes in this mode, we delete all fw_devlink discovered dependencies of that device where the supplier hasn't yet probed successfully because they have to be optional dependencies. This also means that some devices that aren't needed for init and could have waited for their optional supplier to probe (when the supplier's module is loaded later on) would end up probing prematurely with limited functionality. So call this function only when boot would fail without it. Tested-by: Geert Uytterhoeven <geert+renesas@glider.be> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20220601070707.3946847-5-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-06-01 07:07:00 +00:00
int ret = 0, fwnode_ret = 0;
struct fwnode_handle *sup_fw;
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
/*
* Device waiting for supplier to become available is not allowed to
* probe.
*/
mutex_lock(&fwnode_link_lock);
if (dev->fwnode && !list_empty(&dev->fwnode->suppliers) &&
!fw_devlink_is_permissive()) {
sup_fw = list_first_entry(&dev->fwnode->suppliers,
struct fwnode_link,
c_hook)->supplier;
driver core: Add wait_for_init_devices_probe helper function Some devices might need to be probed and bound successfully before the kernel boot sequence can finish and move on to init/userspace. For example, a network interface might need to be bound to be able to mount a NFS rootfs. With fw_devlink=on by default, some of these devices might be blocked from probing because they are waiting on a optional supplier that doesn't have a driver. While fw_devlink will eventually identify such devices and unblock the probing automatically, it might be too late by the time it unblocks the probing of devices. For example, the IP4 autoconfig might timeout before fw_devlink unblocks probing of the network interface. This function is available to temporarily try and probe all devices that have a driver even if some of their suppliers haven't been added or don't have drivers. The drivers can then decide which of the suppliers are optional vs mandatory and probe the device if possible. By the time this function returns, all such "best effort" probes are guaranteed to be completed. If a device successfully probes in this mode, we delete all fw_devlink discovered dependencies of that device where the supplier hasn't yet probed successfully because they have to be optional dependencies. This also means that some devices that aren't needed for init and could have waited for their optional supplier to probe (when the supplier's module is loaded later on) would end up probing prematurely with limited functionality. So call this function only when boot would fail without it. Tested-by: Geert Uytterhoeven <geert+renesas@glider.be> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20220601070707.3946847-5-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-06-01 07:07:00 +00:00
if (!dev_is_best_effort(dev)) {
fwnode_ret = -EPROBE_DEFER;
dev_err_probe(dev, -EPROBE_DEFER,
"wait for supplier %pfwP\n", sup_fw);
} else {
fwnode_ret = -EAGAIN;
}
}
mutex_unlock(&fwnode_link_lock);
driver core: Add wait_for_init_devices_probe helper function Some devices might need to be probed and bound successfully before the kernel boot sequence can finish and move on to init/userspace. For example, a network interface might need to be bound to be able to mount a NFS rootfs. With fw_devlink=on by default, some of these devices might be blocked from probing because they are waiting on a optional supplier that doesn't have a driver. While fw_devlink will eventually identify such devices and unblock the probing automatically, it might be too late by the time it unblocks the probing of devices. For example, the IP4 autoconfig might timeout before fw_devlink unblocks probing of the network interface. This function is available to temporarily try and probe all devices that have a driver even if some of their suppliers haven't been added or don't have drivers. The drivers can then decide which of the suppliers are optional vs mandatory and probe the device if possible. By the time this function returns, all such "best effort" probes are guaranteed to be completed. If a device successfully probes in this mode, we delete all fw_devlink discovered dependencies of that device where the supplier hasn't yet probed successfully because they have to be optional dependencies. This also means that some devices that aren't needed for init and could have waited for their optional supplier to probe (when the supplier's module is loaded later on) would end up probing prematurely with limited functionality. So call this function only when boot would fail without it. Tested-by: Geert Uytterhoeven <geert+renesas@glider.be> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20220601070707.3946847-5-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-06-01 07:07:00 +00:00
if (fwnode_ret == -EPROBE_DEFER)
return fwnode_ret;
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
device_links_write_lock();
list_for_each_entry(link, &dev->links.suppliers, c_node) {
driver core: Update device link status correctly for SYNC_STATE_ONLY links When SYNC_STATE_ONLY support was added in commit 05ef983e0d65 ("driver core: Add device link support for SYNC_STATE_ONLY flag"), SYNC_STATE_ONLY links were treated similar to STATELESS links in terms of not blocking consumer probe if the supplier hasn't probed yet. That caused a SYNC_STATE_ONLY device link's status to not get updated. Since SYNC_STATE_ONLY device link is no longer useful once the consumer probes, commit 21c27f06587d ("driver core: Fix SYNC_STATE_ONLY device link implementation") addresses the status update issue by deleting the SYNC_STATE_ONLY device link instead of complicating the status update code. However, there are still some cases where we need to update the status of a SYNC_STATE_ONLY device link. This is because a SYNC_STATE_ONLY device link can later get converted into a normal MANAGED device link when a normal MANAGED device link is created between a supplier and consumer that already have a SYNC_STATE_ONLY device link between them. If a SYNC_STATE_ONLY device link's status isn't maintained correctly till it's converted to a normal MANAGED device link, then the normal MANAGED device link will end up with a wrong link status. This can cause a warning stack trace[1] when the consumer device probes successfully. This commit fixes the SYNC_STATE_ONLY device link status update issue where it wouldn't transition correctly from DL_STATE_DORMANT or DL_STATE_AVAILABLE to DL_STATE_CONSUMER_PROBE. It also resets the status back to DL_STATE_DORMANT or DL_STATE_AVAILABLE if the consumer probe fails. [1] - https://lore.kernel.org/lkml/20200522204120.3b3c9ed6@apollo/ Fixes: 05ef983e0d65 ("driver core: Add device link support for SYNC_STATE_ONLY flag") Fixes: 21c27f06587d ("driver core: Fix SYNC_STATE_ONLY device link implementation") Reported-by: Michael Walle <michael@walle.cc> Tested-by: Michael Walle <michael@walle.cc> Signed-off-by: Saravana Kannan <saravanak@google.com> Reviewed-by: Rafael J. Wysocki <rrafael.j.wysocki@intel.com> Link: https://lore.kernel.org/r/20200526220928.49939-1-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-05-26 22:09:27 +00:00
if (!(link->flags & DL_FLAG_MANAGED))
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
continue;
driver core: Update device link status correctly for SYNC_STATE_ONLY links When SYNC_STATE_ONLY support was added in commit 05ef983e0d65 ("driver core: Add device link support for SYNC_STATE_ONLY flag"), SYNC_STATE_ONLY links were treated similar to STATELESS links in terms of not blocking consumer probe if the supplier hasn't probed yet. That caused a SYNC_STATE_ONLY device link's status to not get updated. Since SYNC_STATE_ONLY device link is no longer useful once the consumer probes, commit 21c27f06587d ("driver core: Fix SYNC_STATE_ONLY device link implementation") addresses the status update issue by deleting the SYNC_STATE_ONLY device link instead of complicating the status update code. However, there are still some cases where we need to update the status of a SYNC_STATE_ONLY device link. This is because a SYNC_STATE_ONLY device link can later get converted into a normal MANAGED device link when a normal MANAGED device link is created between a supplier and consumer that already have a SYNC_STATE_ONLY device link between them. If a SYNC_STATE_ONLY device link's status isn't maintained correctly till it's converted to a normal MANAGED device link, then the normal MANAGED device link will end up with a wrong link status. This can cause a warning stack trace[1] when the consumer device probes successfully. This commit fixes the SYNC_STATE_ONLY device link status update issue where it wouldn't transition correctly from DL_STATE_DORMANT or DL_STATE_AVAILABLE to DL_STATE_CONSUMER_PROBE. It also resets the status back to DL_STATE_DORMANT or DL_STATE_AVAILABLE if the consumer probe fails. [1] - https://lore.kernel.org/lkml/20200522204120.3b3c9ed6@apollo/ Fixes: 05ef983e0d65 ("driver core: Add device link support for SYNC_STATE_ONLY flag") Fixes: 21c27f06587d ("driver core: Fix SYNC_STATE_ONLY device link implementation") Reported-by: Michael Walle <michael@walle.cc> Tested-by: Michael Walle <michael@walle.cc> Signed-off-by: Saravana Kannan <saravanak@google.com> Reviewed-by: Rafael J. Wysocki <rrafael.j.wysocki@intel.com> Link: https://lore.kernel.org/r/20200526220928.49939-1-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-05-26 22:09:27 +00:00
if (link->status != DL_STATE_AVAILABLE &&
!(link->flags & DL_FLAG_SYNC_STATE_ONLY)) {
driver core: Add wait_for_init_devices_probe helper function Some devices might need to be probed and bound successfully before the kernel boot sequence can finish and move on to init/userspace. For example, a network interface might need to be bound to be able to mount a NFS rootfs. With fw_devlink=on by default, some of these devices might be blocked from probing because they are waiting on a optional supplier that doesn't have a driver. While fw_devlink will eventually identify such devices and unblock the probing automatically, it might be too late by the time it unblocks the probing of devices. For example, the IP4 autoconfig might timeout before fw_devlink unblocks probing of the network interface. This function is available to temporarily try and probe all devices that have a driver even if some of their suppliers haven't been added or don't have drivers. The drivers can then decide which of the suppliers are optional vs mandatory and probe the device if possible. By the time this function returns, all such "best effort" probes are guaranteed to be completed. If a device successfully probes in this mode, we delete all fw_devlink discovered dependencies of that device where the supplier hasn't yet probed successfully because they have to be optional dependencies. This also means that some devices that aren't needed for init and could have waited for their optional supplier to probe (when the supplier's module is loaded later on) would end up probing prematurely with limited functionality. So call this function only when boot would fail without it. Tested-by: Geert Uytterhoeven <geert+renesas@glider.be> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20220601070707.3946847-5-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-06-01 07:07:00 +00:00
if (dev_is_best_effort(dev) &&
link->flags & DL_FLAG_INFERRED &&
!link->supplier->can_match) {
ret = -EAGAIN;
continue;
}
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
device_links_missing_supplier(dev);
dev_err_probe(dev, -EPROBE_DEFER,
"supplier %s not ready\n",
dev_name(link->supplier));
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
ret = -EPROBE_DEFER;
break;
}
WRITE_ONCE(link->status, DL_STATE_CONSUMER_PROBE);
}
dev->links.status = DL_DEV_PROBING;
device_links_write_unlock();
driver core: Add wait_for_init_devices_probe helper function Some devices might need to be probed and bound successfully before the kernel boot sequence can finish and move on to init/userspace. For example, a network interface might need to be bound to be able to mount a NFS rootfs. With fw_devlink=on by default, some of these devices might be blocked from probing because they are waiting on a optional supplier that doesn't have a driver. While fw_devlink will eventually identify such devices and unblock the probing automatically, it might be too late by the time it unblocks the probing of devices. For example, the IP4 autoconfig might timeout before fw_devlink unblocks probing of the network interface. This function is available to temporarily try and probe all devices that have a driver even if some of their suppliers haven't been added or don't have drivers. The drivers can then decide which of the suppliers are optional vs mandatory and probe the device if possible. By the time this function returns, all such "best effort" probes are guaranteed to be completed. If a device successfully probes in this mode, we delete all fw_devlink discovered dependencies of that device where the supplier hasn't yet probed successfully because they have to be optional dependencies. This also means that some devices that aren't needed for init and could have waited for their optional supplier to probe (when the supplier's module is loaded later on) would end up probing prematurely with limited functionality. So call this function only when boot would fail without it. Tested-by: Geert Uytterhoeven <geert+renesas@glider.be> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20220601070707.3946847-5-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-06-01 07:07:00 +00:00
return ret ? ret : fwnode_ret;
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
}
/**
* __device_links_queue_sync_state - Queue a device for sync_state() callback
* @dev: Device to call sync_state() on
* @list: List head to queue the @dev on
*
* Queues a device for a sync_state() callback when the device links write lock
* isn't held. This allows the sync_state() execution flow to use device links
* APIs. The caller must ensure this function is called with
* device_links_write_lock() held.
*
* This function does a get_device() to make sure the device is not freed while
* on this list.
*
* So the caller must also ensure that device_links_flush_sync_list() is called
* as soon as the caller releases device_links_write_lock(). This is necessary
* to make sure the sync_state() is called in a timely fashion and the
* put_device() is called on this device.
*/
static void __device_links_queue_sync_state(struct device *dev,
struct list_head *list)
driver core: Add sync_state driver/bus callback This sync_state driver/bus callback is called once all the consumers of a supplier have probed successfully. This allows the supplier device's driver/bus to sync the supplier device's state to the software state with the guarantee that all the consumers are actively managing the resources provided by the supplier device. To maintain backwards compatibility and ease transition from existing frameworks and resource cleanup schemes, late_initcall_sync is the earliest when the sync_state callback might be called. There is no upper bound on the time by which the sync_state callback has to be called. This is because if a consumer device never probes, the supplier has to maintain its resources in the state left by the bootloader. For example, if the bootloader leaves the display backlight at a fixed voltage and the backlight driver is never probed, you don't want the backlight to ever be turned off after boot up. Also, when multiple devices are added after kernel init, some suppliers could be added before their consumer devices get added. In these instances, the supplier devices could get their sync_state callback called right after they probe because the consumers devices haven't had a chance to create device links to the suppliers. To handle this correctly, this change also provides APIs to pause/resume sync state callbacks so that when multiple devices are added, their sync_state callback evaluation can be postponed to happen after all of them are added. kbuild test robot reported missing documentation for device.state_synced Reported-by: kbuild test robot <lkp@intel.com> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20190904211126.47518-5-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-09-04 21:11:23 +00:00
{
struct device_link *link;
if (!dev_has_sync_state(dev))
return;
driver core: Add sync_state driver/bus callback This sync_state driver/bus callback is called once all the consumers of a supplier have probed successfully. This allows the supplier device's driver/bus to sync the supplier device's state to the software state with the guarantee that all the consumers are actively managing the resources provided by the supplier device. To maintain backwards compatibility and ease transition from existing frameworks and resource cleanup schemes, late_initcall_sync is the earliest when the sync_state callback might be called. There is no upper bound on the time by which the sync_state callback has to be called. This is because if a consumer device never probes, the supplier has to maintain its resources in the state left by the bootloader. For example, if the bootloader leaves the display backlight at a fixed voltage and the backlight driver is never probed, you don't want the backlight to ever be turned off after boot up. Also, when multiple devices are added after kernel init, some suppliers could be added before their consumer devices get added. In these instances, the supplier devices could get their sync_state callback called right after they probe because the consumers devices haven't had a chance to create device links to the suppliers. To handle this correctly, this change also provides APIs to pause/resume sync state callbacks so that when multiple devices are added, their sync_state callback evaluation can be postponed to happen after all of them are added. kbuild test robot reported missing documentation for device.state_synced Reported-by: kbuild test robot <lkp@intel.com> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20190904211126.47518-5-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-09-04 21:11:23 +00:00
if (dev->state_synced)
return;
list_for_each_entry(link, &dev->links.consumers, s_node) {
if (!(link->flags & DL_FLAG_MANAGED))
continue;
if (link->status != DL_STATE_ACTIVE)
return;
}
/*
* Set the flag here to avoid adding the same device to a list more
* than once. This can happen if new consumers get added to the device
* and probed before the list is flushed.
*/
driver core: Add sync_state driver/bus callback This sync_state driver/bus callback is called once all the consumers of a supplier have probed successfully. This allows the supplier device's driver/bus to sync the supplier device's state to the software state with the guarantee that all the consumers are actively managing the resources provided by the supplier device. To maintain backwards compatibility and ease transition from existing frameworks and resource cleanup schemes, late_initcall_sync is the earliest when the sync_state callback might be called. There is no upper bound on the time by which the sync_state callback has to be called. This is because if a consumer device never probes, the supplier has to maintain its resources in the state left by the bootloader. For example, if the bootloader leaves the display backlight at a fixed voltage and the backlight driver is never probed, you don't want the backlight to ever be turned off after boot up. Also, when multiple devices are added after kernel init, some suppliers could be added before their consumer devices get added. In these instances, the supplier devices could get their sync_state callback called right after they probe because the consumers devices haven't had a chance to create device links to the suppliers. To handle this correctly, this change also provides APIs to pause/resume sync state callbacks so that when multiple devices are added, their sync_state callback evaluation can be postponed to happen after all of them are added. kbuild test robot reported missing documentation for device.state_synced Reported-by: kbuild test robot <lkp@intel.com> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20190904211126.47518-5-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-09-04 21:11:23 +00:00
dev->state_synced = true;
if (WARN_ON(!list_empty(&dev->links.defer_sync)))
return;
get_device(dev);
list_add_tail(&dev->links.defer_sync, list);
}
/**
* device_links_flush_sync_list - Call sync_state() on a list of devices
* @list: List of devices to call sync_state() on
* @dont_lock_dev: Device for which lock is already held by the caller
*
* Calls sync_state() on all the devices that have been queued for it. This
* function is used in conjunction with __device_links_queue_sync_state(). The
* @dont_lock_dev parameter is useful when this function is called from a
* context where a device lock is already held.
*/
static void device_links_flush_sync_list(struct list_head *list,
struct device *dont_lock_dev)
{
struct device *dev, *tmp;
list_for_each_entry_safe(dev, tmp, list, links.defer_sync) {
list_del_init(&dev->links.defer_sync);
if (dev != dont_lock_dev)
device_lock(dev);
if (dev->bus->sync_state)
dev->bus->sync_state(dev);
else if (dev->driver && dev->driver->sync_state)
dev->driver->sync_state(dev);
if (dev != dont_lock_dev)
device_unlock(dev);
put_device(dev);
}
driver core: Add sync_state driver/bus callback This sync_state driver/bus callback is called once all the consumers of a supplier have probed successfully. This allows the supplier device's driver/bus to sync the supplier device's state to the software state with the guarantee that all the consumers are actively managing the resources provided by the supplier device. To maintain backwards compatibility and ease transition from existing frameworks and resource cleanup schemes, late_initcall_sync is the earliest when the sync_state callback might be called. There is no upper bound on the time by which the sync_state callback has to be called. This is because if a consumer device never probes, the supplier has to maintain its resources in the state left by the bootloader. For example, if the bootloader leaves the display backlight at a fixed voltage and the backlight driver is never probed, you don't want the backlight to ever be turned off after boot up. Also, when multiple devices are added after kernel init, some suppliers could be added before their consumer devices get added. In these instances, the supplier devices could get their sync_state callback called right after they probe because the consumers devices haven't had a chance to create device links to the suppliers. To handle this correctly, this change also provides APIs to pause/resume sync state callbacks so that when multiple devices are added, their sync_state callback evaluation can be postponed to happen after all of them are added. kbuild test robot reported missing documentation for device.state_synced Reported-by: kbuild test robot <lkp@intel.com> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20190904211126.47518-5-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-09-04 21:11:23 +00:00
}
void device_links_supplier_sync_state_pause(void)
{
device_links_write_lock();
defer_sync_state_count++;
device_links_write_unlock();
}
void device_links_supplier_sync_state_resume(void)
{
struct device *dev, *tmp;
LIST_HEAD(sync_list);
driver core: Add sync_state driver/bus callback This sync_state driver/bus callback is called once all the consumers of a supplier have probed successfully. This allows the supplier device's driver/bus to sync the supplier device's state to the software state with the guarantee that all the consumers are actively managing the resources provided by the supplier device. To maintain backwards compatibility and ease transition from existing frameworks and resource cleanup schemes, late_initcall_sync is the earliest when the sync_state callback might be called. There is no upper bound on the time by which the sync_state callback has to be called. This is because if a consumer device never probes, the supplier has to maintain its resources in the state left by the bootloader. For example, if the bootloader leaves the display backlight at a fixed voltage and the backlight driver is never probed, you don't want the backlight to ever be turned off after boot up. Also, when multiple devices are added after kernel init, some suppliers could be added before their consumer devices get added. In these instances, the supplier devices could get their sync_state callback called right after they probe because the consumers devices haven't had a chance to create device links to the suppliers. To handle this correctly, this change also provides APIs to pause/resume sync state callbacks so that when multiple devices are added, their sync_state callback evaluation can be postponed to happen after all of them are added. kbuild test robot reported missing documentation for device.state_synced Reported-by: kbuild test robot <lkp@intel.com> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20190904211126.47518-5-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-09-04 21:11:23 +00:00
device_links_write_lock();
if (!defer_sync_state_count) {
WARN(true, "Unmatched sync_state pause/resume!");
goto out;
}
defer_sync_state_count--;
if (defer_sync_state_count)
goto out;
list_for_each_entry_safe(dev, tmp, &deferred_sync, links.defer_sync) {
/*
* Delete from deferred_sync list before queuing it to
* sync_list because defer_sync is used for both lists.
*/
list_del_init(&dev->links.defer_sync);
__device_links_queue_sync_state(dev, &sync_list);
driver core: Add sync_state driver/bus callback This sync_state driver/bus callback is called once all the consumers of a supplier have probed successfully. This allows the supplier device's driver/bus to sync the supplier device's state to the software state with the guarantee that all the consumers are actively managing the resources provided by the supplier device. To maintain backwards compatibility and ease transition from existing frameworks and resource cleanup schemes, late_initcall_sync is the earliest when the sync_state callback might be called. There is no upper bound on the time by which the sync_state callback has to be called. This is because if a consumer device never probes, the supplier has to maintain its resources in the state left by the bootloader. For example, if the bootloader leaves the display backlight at a fixed voltage and the backlight driver is never probed, you don't want the backlight to ever be turned off after boot up. Also, when multiple devices are added after kernel init, some suppliers could be added before their consumer devices get added. In these instances, the supplier devices could get their sync_state callback called right after they probe because the consumers devices haven't had a chance to create device links to the suppliers. To handle this correctly, this change also provides APIs to pause/resume sync state callbacks so that when multiple devices are added, their sync_state callback evaluation can be postponed to happen after all of them are added. kbuild test robot reported missing documentation for device.state_synced Reported-by: kbuild test robot <lkp@intel.com> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20190904211126.47518-5-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-09-04 21:11:23 +00:00
}
out:
device_links_write_unlock();
device_links_flush_sync_list(&sync_list, NULL);
driver core: Add sync_state driver/bus callback This sync_state driver/bus callback is called once all the consumers of a supplier have probed successfully. This allows the supplier device's driver/bus to sync the supplier device's state to the software state with the guarantee that all the consumers are actively managing the resources provided by the supplier device. To maintain backwards compatibility and ease transition from existing frameworks and resource cleanup schemes, late_initcall_sync is the earliest when the sync_state callback might be called. There is no upper bound on the time by which the sync_state callback has to be called. This is because if a consumer device never probes, the supplier has to maintain its resources in the state left by the bootloader. For example, if the bootloader leaves the display backlight at a fixed voltage and the backlight driver is never probed, you don't want the backlight to ever be turned off after boot up. Also, when multiple devices are added after kernel init, some suppliers could be added before their consumer devices get added. In these instances, the supplier devices could get their sync_state callback called right after they probe because the consumers devices haven't had a chance to create device links to the suppliers. To handle this correctly, this change also provides APIs to pause/resume sync state callbacks so that when multiple devices are added, their sync_state callback evaluation can be postponed to happen after all of them are added. kbuild test robot reported missing documentation for device.state_synced Reported-by: kbuild test robot <lkp@intel.com> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20190904211126.47518-5-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-09-04 21:11:23 +00:00
}
static int sync_state_resume_initcall(void)
{
device_links_supplier_sync_state_resume();
return 0;
}
late_initcall(sync_state_resume_initcall);
static void __device_links_supplier_defer_sync(struct device *sup)
{
if (list_empty(&sup->links.defer_sync) && dev_has_sync_state(sup))
list_add_tail(&sup->links.defer_sync, &deferred_sync);
driver core: Add sync_state driver/bus callback This sync_state driver/bus callback is called once all the consumers of a supplier have probed successfully. This allows the supplier device's driver/bus to sync the supplier device's state to the software state with the guarantee that all the consumers are actively managing the resources provided by the supplier device. To maintain backwards compatibility and ease transition from existing frameworks and resource cleanup schemes, late_initcall_sync is the earliest when the sync_state callback might be called. There is no upper bound on the time by which the sync_state callback has to be called. This is because if a consumer device never probes, the supplier has to maintain its resources in the state left by the bootloader. For example, if the bootloader leaves the display backlight at a fixed voltage and the backlight driver is never probed, you don't want the backlight to ever be turned off after boot up. Also, when multiple devices are added after kernel init, some suppliers could be added before their consumer devices get added. In these instances, the supplier devices could get their sync_state callback called right after they probe because the consumers devices haven't had a chance to create device links to the suppliers. To handle this correctly, this change also provides APIs to pause/resume sync state callbacks so that when multiple devices are added, their sync_state callback evaluation can be postponed to happen after all of them are added. kbuild test robot reported missing documentation for device.state_synced Reported-by: kbuild test robot <lkp@intel.com> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20190904211126.47518-5-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-09-04 21:11:23 +00:00
}
driver core: Fix SYNC_STATE_ONLY device link implementation When SYNC_STATE_ONLY support was added in commit 05ef983e0d65 ("driver core: Add device link support for SYNC_STATE_ONLY flag"), device_link_add() incorrectly skipped adding the new SYNC_STATE_ONLY device link to the supplier's and consumer's "device link" list. This causes multiple issues: - The device link is lost forever from driver core if the caller didn't keep track of it (caller typically isn't expected to). This is a memory leak. - The device link is also never visible to any other code path after device_link_add() returns. If we fix the "device link" list handling, that exposes a bunch of issues. 1. The device link "status" state management code rightfully doesn't handle the case where a DL_FLAG_MANAGED device link exists between a supplier and consumer, but the consumer manages to probe successfully before the supplier. The addition of DL_FLAG_SYNC_STATE_ONLY links break this assumption. This causes device_links_driver_bound() to throw a warning when this happens. Since DL_FLAG_SYNC_STATE_ONLY device links are mainly used for creating proxy device links for child device dependencies and aren't useful once the consumer device probes successfully, this patch just deletes DL_FLAG_SYNC_STATE_ONLY device links once its consumer device probes. This way, we avoid the warning, free up some memory and avoid complicating the device links "status" state management code. 2. Creating a DL_FLAG_STATELESS device link between two devices that already have a DL_FLAG_SYNC_STATE_ONLY device link will result in the DL_FLAG_STATELESS flag not getting set correctly. This patch also fixes this. Lastly, this patch also fixes minor whitespace issues. Cc: stable@vger.kernel.org Fixes: 05ef983e0d65 ("driver core: Add device link support for SYNC_STATE_ONLY flag") Signed-off-by: Saravana Kannan <saravanak@google.com> Reviewed-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Link: https://lore.kernel.org/r/20200519063000.128819-1-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-05-19 06:30:00 +00:00
static void device_link_drop_managed(struct device_link *link)
{
link->flags &= ~DL_FLAG_MANAGED;
WRITE_ONCE(link->status, DL_STATE_NONE);
kref_put(&link->kref, __device_link_del);
}
static ssize_t waiting_for_supplier_show(struct device *dev,
struct device_attribute *attr,
char *buf)
{
bool val;
device_lock(dev);
val = !list_empty(&dev->fwnode->suppliers);
device_unlock(dev);
drivers core: Use sysfs_emit and sysfs_emit_at for show(device *...) functions Convert the various sprintf fmaily calls in sysfs device show functions to sysfs_emit and sysfs_emit_at for PAGE_SIZE buffer safety. Done with: $ spatch -sp-file sysfs_emit_dev.cocci --in-place --max-width=80 . And cocci script: $ cat sysfs_emit_dev.cocci @@ identifier d_show; identifier dev, attr, buf; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... return - sprintf(buf, + sysfs_emit(buf, ...); ...> } @@ identifier d_show; identifier dev, attr, buf; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... return - snprintf(buf, PAGE_SIZE, + sysfs_emit(buf, ...); ...> } @@ identifier d_show; identifier dev, attr, buf; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... return - scnprintf(buf, PAGE_SIZE, + sysfs_emit(buf, ...); ...> } @@ identifier d_show; identifier dev, attr, buf; expression chr; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... return - strcpy(buf, chr); + sysfs_emit(buf, chr); ...> } @@ identifier d_show; identifier dev, attr, buf; identifier len; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... len = - sprintf(buf, + sysfs_emit(buf, ...); ...> return len; } @@ identifier d_show; identifier dev, attr, buf; identifier len; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... len = - snprintf(buf, PAGE_SIZE, + sysfs_emit(buf, ...); ...> return len; } @@ identifier d_show; identifier dev, attr, buf; identifier len; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... len = - scnprintf(buf, PAGE_SIZE, + sysfs_emit(buf, ...); ...> return len; } @@ identifier d_show; identifier dev, attr, buf; identifier len; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... - len += scnprintf(buf + len, PAGE_SIZE - len, + len += sysfs_emit_at(buf, len, ...); ...> return len; } @@ identifier d_show; identifier dev, attr, buf; expression chr; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { ... - strcpy(buf, chr); - return strlen(buf); + return sysfs_emit(buf, chr); } Signed-off-by: Joe Perches <joe@perches.com> Link: https://lore.kernel.org/r/3d033c33056d88bbe34d4ddb62afd05ee166ab9a.1600285923.git.joe@perches.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-09-16 20:40:39 +00:00
return sysfs_emit(buf, "%u\n", val);
}
static DEVICE_ATTR_RO(waiting_for_supplier);
/**
* device_links_force_bind - Prepares device to be force bound
* @dev: Consumer device.
*
* device_bind_driver() force binds a device to a driver without calling any
* driver probe functions. So the consumer really isn't going to wait for any
* supplier before it's bound to the driver. We still want the device link
* states to be sensible when this happens.
*
* In preparation for device_bind_driver(), this function goes through each
* supplier device links and checks if the supplier is bound. If it is, then
* the device link status is set to CONSUMER_PROBE. Otherwise, the device link
* is dropped. Links without the DL_FLAG_MANAGED flag set are ignored.
*/
void device_links_force_bind(struct device *dev)
{
struct device_link *link, *ln;
device_links_write_lock();
list_for_each_entry_safe(link, ln, &dev->links.suppliers, c_node) {
if (!(link->flags & DL_FLAG_MANAGED))
continue;
if (link->status != DL_STATE_AVAILABLE) {
device_link_drop_managed(link);
continue;
}
WRITE_ONCE(link->status, DL_STATE_CONSUMER_PROBE);
}
dev->links.status = DL_DEV_PROBING;
device_links_write_unlock();
}
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
/**
* device_links_driver_bound - Update device links after probing its driver.
* @dev: Device to update the links for.
*
* The probe has been successful, so update links from this device to any
* consumers by changing their status to "available".
*
* Also change the status of @dev's links to suppliers to "active".
*
driver core: Remove device link creation limitation If device_link_add() is called for a consumer/supplier pair with an existing device link between them and the existing link's type is not in agreement with the flags passed to that function by its caller, NULL will be returned. That is seriously inconvenient, because it forces the callers of device_link_add() to worry about what others may or may not do even if that is not relevant to them for any other reasons. It turns out, however, that this limitation can be made go away relatively easily. The underlying observation is that if DL_FLAG_STATELESS has been passed to device_link_add() in flags for the given consumer/supplier pair at least once, calling either device_link_del() or device_link_remove() to release the link returned by it should work, but there are no other requirements associated with that flag. In turn, if at least one of the callers of device_link_add() for the given consumer/supplier pair has not passed DL_FLAG_STATELESS to it in flags, the driver core should track the status of the link and act on it as appropriate (ie. the link should be treated as "managed"). This means that DL_FLAG_STATELESS needs to be set for managed device links and it should be valid to call device_link_del() or device_link_remove() to drop references to them in certain sutiations. To allow that to happen, introduce a new (internal) device link flag called DL_FLAG_MANAGED and make device_link_add() set it automatically whenever DL_FLAG_STATELESS is not passed to it. Also make it take additional references to existing device links that were previously stateless (that is, with DL_FLAG_STATELESS set and DL_FLAG_MANAGED unset) and will need to be managed going forward and initialize their status (which has been DL_STATE_NONE so far). Accordingly, when a managed device link is dropped automatically by the driver core, make it clear DL_FLAG_MANAGED, reset the link's status back to DL_STATE_NONE and drop the reference to it associated with DL_FLAG_MANAGED instead of just deleting it right away (to allow it to stay around in case it still needs to be released explicitly by someone). With that, since setting DL_FLAG_STATELESS doesn't mean that the device link in question is not managed any more, replace all of the status-tracking checks against DL_FLAG_STATELESS with analogous checks against DL_FLAG_MANAGED and update the documentation to reflect these changes. While at it, make device_link_add() reject flags that it does not recognize, including DL_FLAG_MANAGED. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Reviewed-by: Saravana Kannan <saravanak@google.com> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Review-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/2305283.AStDPdUUnE@kreacher Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-07-16 15:21:06 +00:00
* Links without the DL_FLAG_MANAGED flag set are ignored.
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
*/
void device_links_driver_bound(struct device *dev)
{
driver core: Fix SYNC_STATE_ONLY device link implementation When SYNC_STATE_ONLY support was added in commit 05ef983e0d65 ("driver core: Add device link support for SYNC_STATE_ONLY flag"), device_link_add() incorrectly skipped adding the new SYNC_STATE_ONLY device link to the supplier's and consumer's "device link" list. This causes multiple issues: - The device link is lost forever from driver core if the caller didn't keep track of it (caller typically isn't expected to). This is a memory leak. - The device link is also never visible to any other code path after device_link_add() returns. If we fix the "device link" list handling, that exposes a bunch of issues. 1. The device link "status" state management code rightfully doesn't handle the case where a DL_FLAG_MANAGED device link exists between a supplier and consumer, but the consumer manages to probe successfully before the supplier. The addition of DL_FLAG_SYNC_STATE_ONLY links break this assumption. This causes device_links_driver_bound() to throw a warning when this happens. Since DL_FLAG_SYNC_STATE_ONLY device links are mainly used for creating proxy device links for child device dependencies and aren't useful once the consumer device probes successfully, this patch just deletes DL_FLAG_SYNC_STATE_ONLY device links once its consumer device probes. This way, we avoid the warning, free up some memory and avoid complicating the device links "status" state management code. 2. Creating a DL_FLAG_STATELESS device link between two devices that already have a DL_FLAG_SYNC_STATE_ONLY device link will result in the DL_FLAG_STATELESS flag not getting set correctly. This patch also fixes this. Lastly, this patch also fixes minor whitespace issues. Cc: stable@vger.kernel.org Fixes: 05ef983e0d65 ("driver core: Add device link support for SYNC_STATE_ONLY flag") Signed-off-by: Saravana Kannan <saravanak@google.com> Reviewed-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Link: https://lore.kernel.org/r/20200519063000.128819-1-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-05-19 06:30:00 +00:00
struct device_link *link, *ln;
LIST_HEAD(sync_list);
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
/*
driver core: fw_devlink: Detect supplier devices that will never be added During the initial parsing of firmware by fw_devlink, fw_devlink might infer that some supplier firmware nodes would get populated as devices. But the inference is not always correct. This patch tries to logically detect and fix such mistakes as boot progresses or more devices probe. fw_devlink makes a fundamental assumption that once a device binds to a driver, it will populate (i.e: add as struct devices) all the child firmware nodes that could be populated as devices (if they aren't populated already). So, whenever a device probes, we check all its child firmware nodes. If a child firmware node has a corresponding device populated, we don't modify the child node or its descendants. However, if a child firmware node has not been populated as a device, we delete all the fwnode links where the child node or its descendants are suppliers. This ensures that no other device is blocked on a firmware node that will never be populated as a device. We also mark such fwnodes as NOT_DEVICE, so that no new fwnode links are created with these nodes as suppliers. Fixes: e590474768f1 ("driver core: Set fw_devlink=on by default") Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Acked-by: Rafael J. Wysocki <rafael@kernel.org> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20210205222644.2357303-2-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2021-02-05 22:26:37 +00:00
* If a device binds successfully, it's expected to have created all
* the device links it needs to or make new device links as it needs
driver core: fw_devlink: Detect supplier devices that will never be added During the initial parsing of firmware by fw_devlink, fw_devlink might infer that some supplier firmware nodes would get populated as devices. But the inference is not always correct. This patch tries to logically detect and fix such mistakes as boot progresses or more devices probe. fw_devlink makes a fundamental assumption that once a device binds to a driver, it will populate (i.e: add as struct devices) all the child firmware nodes that could be populated as devices (if they aren't populated already). So, whenever a device probes, we check all its child firmware nodes. If a child firmware node has a corresponding device populated, we don't modify the child node or its descendants. However, if a child firmware node has not been populated as a device, we delete all the fwnode links where the child node or its descendants are suppliers. This ensures that no other device is blocked on a firmware node that will never be populated as a device. We also mark such fwnodes as NOT_DEVICE, so that no new fwnode links are created with these nodes as suppliers. Fixes: e590474768f1 ("driver core: Set fw_devlink=on by default") Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Acked-by: Rafael J. Wysocki <rafael@kernel.org> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20210205222644.2357303-2-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2021-02-05 22:26:37 +00:00
* them. So, fw_devlink no longer needs to create device links to any
* of the device's suppliers.
*
* Also, if a child firmware node of this bound device is not added as
* a device by now, assume it is never going to be added and make sure
* other devices don't defer probe indefinitely by waiting for such a
* child device.
*/
driver core: fw_devlink: Detect supplier devices that will never be added During the initial parsing of firmware by fw_devlink, fw_devlink might infer that some supplier firmware nodes would get populated as devices. But the inference is not always correct. This patch tries to logically detect and fix such mistakes as boot progresses or more devices probe. fw_devlink makes a fundamental assumption that once a device binds to a driver, it will populate (i.e: add as struct devices) all the child firmware nodes that could be populated as devices (if they aren't populated already). So, whenever a device probes, we check all its child firmware nodes. If a child firmware node has a corresponding device populated, we don't modify the child node or its descendants. However, if a child firmware node has not been populated as a device, we delete all the fwnode links where the child node or its descendants are suppliers. This ensures that no other device is blocked on a firmware node that will never be populated as a device. We also mark such fwnodes as NOT_DEVICE, so that no new fwnode links are created with these nodes as suppliers. Fixes: e590474768f1 ("driver core: Set fw_devlink=on by default") Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Acked-by: Rafael J. Wysocki <rafael@kernel.org> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20210205222644.2357303-2-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2021-02-05 22:26:37 +00:00
if (dev->fwnode && dev->fwnode->dev == dev) {
struct fwnode_handle *child;
driver core: Refactor fw_devlink feature The current implementation of fw_devlink is very inefficient because it tries to get away without creating fwnode links in the name of saving memory usage. Past attempts to optimize runtime at the cost of memory usage were blocked with request for data showing that the optimization made significant improvement for real world scenarios. We have those scenarios now. There have been several reports of boot time increase in the order of seconds in this thread [1]. Several OEMs and SoC manufacturers have also privately reported significant (350-400ms) increase in boot time due to all the parsing done by fw_devlink. So this patch uses all the setup done by the previous patches in this series to refactor fw_devlink to be more efficient. Most of the code has been moved out of firmware specific (DT mostly) code into driver core. This brings the following benefits: - Instead of parsing the device tree multiple times during bootup, fw_devlink parses each fwnode node/property only once and creates fwnode links. The rest of the fw_devlink code then just looks at these fwnode links to do rest of the work. - Makes it much easier to debug probe issue due to fw_devlink in the future. fw_devlink=on blocks the probing of devices if they depend on a device that hasn't been added yet. With this refactor, it'll be very easy to tell what that device is because we now have a reference to the fwnode of the device. - Much easier to add fw_devlink support to ACPI and other firmware types. A refactor to move the common bits from DT specific code to driver core was in my TODO list as a prerequisite to adding ACPI support to fw_devlink. This series gets that done. [1] - https://lore.kernel.org/linux-omap/ea02f57e-871d-cd16-4418-c1da4bbc4696@ti.com/ Tested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Tested-by: Grygorii Strashko <grygorii.strashko@ti.com> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20201121020232.908850-17-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-11-21 02:02:31 +00:00
fwnode_links_purge_suppliers(dev->fwnode);
driver core: fw_devlink: Detect supplier devices that will never be added During the initial parsing of firmware by fw_devlink, fw_devlink might infer that some supplier firmware nodes would get populated as devices. But the inference is not always correct. This patch tries to logically detect and fix such mistakes as boot progresses or more devices probe. fw_devlink makes a fundamental assumption that once a device binds to a driver, it will populate (i.e: add as struct devices) all the child firmware nodes that could be populated as devices (if they aren't populated already). So, whenever a device probes, we check all its child firmware nodes. If a child firmware node has a corresponding device populated, we don't modify the child node or its descendants. However, if a child firmware node has not been populated as a device, we delete all the fwnode links where the child node or its descendants are suppliers. This ensures that no other device is blocked on a firmware node that will never be populated as a device. We also mark such fwnodes as NOT_DEVICE, so that no new fwnode links are created with these nodes as suppliers. Fixes: e590474768f1 ("driver core: Set fw_devlink=on by default") Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Acked-by: Rafael J. Wysocki <rafael@kernel.org> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20210205222644.2357303-2-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2021-02-05 22:26:37 +00:00
fwnode_for_each_available_child_node(dev->fwnode, child)
fw_devlink_purge_absent_suppliers(child);
}
device_remove_file(dev, &dev_attr_waiting_for_supplier);
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
device_links_write_lock();
list_for_each_entry(link, &dev->links.consumers, s_node) {
driver core: Remove device link creation limitation If device_link_add() is called for a consumer/supplier pair with an existing device link between them and the existing link's type is not in agreement with the flags passed to that function by its caller, NULL will be returned. That is seriously inconvenient, because it forces the callers of device_link_add() to worry about what others may or may not do even if that is not relevant to them for any other reasons. It turns out, however, that this limitation can be made go away relatively easily. The underlying observation is that if DL_FLAG_STATELESS has been passed to device_link_add() in flags for the given consumer/supplier pair at least once, calling either device_link_del() or device_link_remove() to release the link returned by it should work, but there are no other requirements associated with that flag. In turn, if at least one of the callers of device_link_add() for the given consumer/supplier pair has not passed DL_FLAG_STATELESS to it in flags, the driver core should track the status of the link and act on it as appropriate (ie. the link should be treated as "managed"). This means that DL_FLAG_STATELESS needs to be set for managed device links and it should be valid to call device_link_del() or device_link_remove() to drop references to them in certain sutiations. To allow that to happen, introduce a new (internal) device link flag called DL_FLAG_MANAGED and make device_link_add() set it automatically whenever DL_FLAG_STATELESS is not passed to it. Also make it take additional references to existing device links that were previously stateless (that is, with DL_FLAG_STATELESS set and DL_FLAG_MANAGED unset) and will need to be managed going forward and initialize their status (which has been DL_STATE_NONE so far). Accordingly, when a managed device link is dropped automatically by the driver core, make it clear DL_FLAG_MANAGED, reset the link's status back to DL_STATE_NONE and drop the reference to it associated with DL_FLAG_MANAGED instead of just deleting it right away (to allow it to stay around in case it still needs to be released explicitly by someone). With that, since setting DL_FLAG_STATELESS doesn't mean that the device link in question is not managed any more, replace all of the status-tracking checks against DL_FLAG_STATELESS with analogous checks against DL_FLAG_MANAGED and update the documentation to reflect these changes. While at it, make device_link_add() reject flags that it does not recognize, including DL_FLAG_MANAGED. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Reviewed-by: Saravana Kannan <saravanak@google.com> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Review-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/2305283.AStDPdUUnE@kreacher Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-07-16 15:21:06 +00:00
if (!(link->flags & DL_FLAG_MANAGED))
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
continue;
driver core: Fix adding device links to probing suppliers Currently, it is not valid to add a device link from a consumer driver ->probe callback to a supplier that is still probing too, but generally this is a valid use case. For example, if the consumer has just acquired a resource that can only be available if the supplier is functional, adding a device link to that supplier right away should be safe (and even desirable arguably), but device_link_add() doesn't handle that case correctly and the initial state of the link created by it is wrong then. To address this problem, change the initial state of device links added between a probing supplier and a probing consumer to DL_STATE_CONSUMER_PROBE and update device_links_driver_bound() to skip such links on the supplier side. With this change, if the supplier probe completes first, device_links_driver_bound() called for it will skip the link state update and when it is called for the consumer, the link state will be updated to "active". In turn, if the consumer probe completes first, device_links_driver_bound() called for it will change the state of the link to "active" and when it is called for the supplier, the link status update will be skipped. However, in principle the supplier or consumer probe may still fail after the link has been added, so modify device_links_no_driver() to change device links in the "active" or "consumer probe" state to "dormant" on the supplier side and update __device_links_no_driver() to change the link state to "available" only if it is "consumer probe" or "active". Then, if the supplier probe fails first, the leftover link to the probing consumer will become "dormant" and device_links_no_driver() called for the consumer (when its probe fails) will clean it up. In turn, if the consumer probe fails first, it will either drop the link, or change its state to "available" and, in the latter case, when device_links_no_driver() is called for the supplier, it will update the link state to "dormant". [If the supplier probe fails, but the consumer probe succeeds, which should not happen as long as the consumer driver is correct, the link still will be around, but it will be "dormant" until the supplier is probed again.] Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-02-01 00:50:39 +00:00
/*
* Links created during consumer probe may be in the "consumer
* probe" state to start with if the supplier is still probing
* when they are created and they may become "active" if the
* consumer probe returns first. Skip them here.
*/
if (link->status == DL_STATE_CONSUMER_PROBE ||
link->status == DL_STATE_ACTIVE)
continue;
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
WARN_ON(link->status != DL_STATE_DORMANT);
WRITE_ONCE(link->status, DL_STATE_AVAILABLE);
if (link->flags & DL_FLAG_AUTOPROBE_CONSUMER)
driver_deferred_probe_add(link->consumer);
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
}
if (defer_sync_state_count)
__device_links_supplier_defer_sync(dev);
else
__device_links_queue_sync_state(dev, &sync_list);
driver core: Fix SYNC_STATE_ONLY device link implementation When SYNC_STATE_ONLY support was added in commit 05ef983e0d65 ("driver core: Add device link support for SYNC_STATE_ONLY flag"), device_link_add() incorrectly skipped adding the new SYNC_STATE_ONLY device link to the supplier's and consumer's "device link" list. This causes multiple issues: - The device link is lost forever from driver core if the caller didn't keep track of it (caller typically isn't expected to). This is a memory leak. - The device link is also never visible to any other code path after device_link_add() returns. If we fix the "device link" list handling, that exposes a bunch of issues. 1. The device link "status" state management code rightfully doesn't handle the case where a DL_FLAG_MANAGED device link exists between a supplier and consumer, but the consumer manages to probe successfully before the supplier. The addition of DL_FLAG_SYNC_STATE_ONLY links break this assumption. This causes device_links_driver_bound() to throw a warning when this happens. Since DL_FLAG_SYNC_STATE_ONLY device links are mainly used for creating proxy device links for child device dependencies and aren't useful once the consumer device probes successfully, this patch just deletes DL_FLAG_SYNC_STATE_ONLY device links once its consumer device probes. This way, we avoid the warning, free up some memory and avoid complicating the device links "status" state management code. 2. Creating a DL_FLAG_STATELESS device link between two devices that already have a DL_FLAG_SYNC_STATE_ONLY device link will result in the DL_FLAG_STATELESS flag not getting set correctly. This patch also fixes this. Lastly, this patch also fixes minor whitespace issues. Cc: stable@vger.kernel.org Fixes: 05ef983e0d65 ("driver core: Add device link support for SYNC_STATE_ONLY flag") Signed-off-by: Saravana Kannan <saravanak@google.com> Reviewed-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Link: https://lore.kernel.org/r/20200519063000.128819-1-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-05-19 06:30:00 +00:00
list_for_each_entry_safe(link, ln, &dev->links.suppliers, c_node) {
struct device *supplier;
driver core: Remove device link creation limitation If device_link_add() is called for a consumer/supplier pair with an existing device link between them and the existing link's type is not in agreement with the flags passed to that function by its caller, NULL will be returned. That is seriously inconvenient, because it forces the callers of device_link_add() to worry about what others may or may not do even if that is not relevant to them for any other reasons. It turns out, however, that this limitation can be made go away relatively easily. The underlying observation is that if DL_FLAG_STATELESS has been passed to device_link_add() in flags for the given consumer/supplier pair at least once, calling either device_link_del() or device_link_remove() to release the link returned by it should work, but there are no other requirements associated with that flag. In turn, if at least one of the callers of device_link_add() for the given consumer/supplier pair has not passed DL_FLAG_STATELESS to it in flags, the driver core should track the status of the link and act on it as appropriate (ie. the link should be treated as "managed"). This means that DL_FLAG_STATELESS needs to be set for managed device links and it should be valid to call device_link_del() or device_link_remove() to drop references to them in certain sutiations. To allow that to happen, introduce a new (internal) device link flag called DL_FLAG_MANAGED and make device_link_add() set it automatically whenever DL_FLAG_STATELESS is not passed to it. Also make it take additional references to existing device links that were previously stateless (that is, with DL_FLAG_STATELESS set and DL_FLAG_MANAGED unset) and will need to be managed going forward and initialize their status (which has been DL_STATE_NONE so far). Accordingly, when a managed device link is dropped automatically by the driver core, make it clear DL_FLAG_MANAGED, reset the link's status back to DL_STATE_NONE and drop the reference to it associated with DL_FLAG_MANAGED instead of just deleting it right away (to allow it to stay around in case it still needs to be released explicitly by someone). With that, since setting DL_FLAG_STATELESS doesn't mean that the device link in question is not managed any more, replace all of the status-tracking checks against DL_FLAG_STATELESS with analogous checks against DL_FLAG_MANAGED and update the documentation to reflect these changes. While at it, make device_link_add() reject flags that it does not recognize, including DL_FLAG_MANAGED. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Reviewed-by: Saravana Kannan <saravanak@google.com> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Review-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/2305283.AStDPdUUnE@kreacher Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-07-16 15:21:06 +00:00
if (!(link->flags & DL_FLAG_MANAGED))
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
continue;
driver core: Fix SYNC_STATE_ONLY device link implementation When SYNC_STATE_ONLY support was added in commit 05ef983e0d65 ("driver core: Add device link support for SYNC_STATE_ONLY flag"), device_link_add() incorrectly skipped adding the new SYNC_STATE_ONLY device link to the supplier's and consumer's "device link" list. This causes multiple issues: - The device link is lost forever from driver core if the caller didn't keep track of it (caller typically isn't expected to). This is a memory leak. - The device link is also never visible to any other code path after device_link_add() returns. If we fix the "device link" list handling, that exposes a bunch of issues. 1. The device link "status" state management code rightfully doesn't handle the case where a DL_FLAG_MANAGED device link exists between a supplier and consumer, but the consumer manages to probe successfully before the supplier. The addition of DL_FLAG_SYNC_STATE_ONLY links break this assumption. This causes device_links_driver_bound() to throw a warning when this happens. Since DL_FLAG_SYNC_STATE_ONLY device links are mainly used for creating proxy device links for child device dependencies and aren't useful once the consumer device probes successfully, this patch just deletes DL_FLAG_SYNC_STATE_ONLY device links once its consumer device probes. This way, we avoid the warning, free up some memory and avoid complicating the device links "status" state management code. 2. Creating a DL_FLAG_STATELESS device link between two devices that already have a DL_FLAG_SYNC_STATE_ONLY device link will result in the DL_FLAG_STATELESS flag not getting set correctly. This patch also fixes this. Lastly, this patch also fixes minor whitespace issues. Cc: stable@vger.kernel.org Fixes: 05ef983e0d65 ("driver core: Add device link support for SYNC_STATE_ONLY flag") Signed-off-by: Saravana Kannan <saravanak@google.com> Reviewed-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Link: https://lore.kernel.org/r/20200519063000.128819-1-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-05-19 06:30:00 +00:00
supplier = link->supplier;
if (link->flags & DL_FLAG_SYNC_STATE_ONLY) {
/*
* When DL_FLAG_SYNC_STATE_ONLY is set, it means no
* other DL_MANAGED_LINK_FLAGS have been set. So, it's
* save to drop the managed link completely.
*/
device_link_drop_managed(link);
driver core: Add wait_for_init_devices_probe helper function Some devices might need to be probed and bound successfully before the kernel boot sequence can finish and move on to init/userspace. For example, a network interface might need to be bound to be able to mount a NFS rootfs. With fw_devlink=on by default, some of these devices might be blocked from probing because they are waiting on a optional supplier that doesn't have a driver. While fw_devlink will eventually identify such devices and unblock the probing automatically, it might be too late by the time it unblocks the probing of devices. For example, the IP4 autoconfig might timeout before fw_devlink unblocks probing of the network interface. This function is available to temporarily try and probe all devices that have a driver even if some of their suppliers haven't been added or don't have drivers. The drivers can then decide which of the suppliers are optional vs mandatory and probe the device if possible. By the time this function returns, all such "best effort" probes are guaranteed to be completed. If a device successfully probes in this mode, we delete all fw_devlink discovered dependencies of that device where the supplier hasn't yet probed successfully because they have to be optional dependencies. This also means that some devices that aren't needed for init and could have waited for their optional supplier to probe (when the supplier's module is loaded later on) would end up probing prematurely with limited functionality. So call this function only when boot would fail without it. Tested-by: Geert Uytterhoeven <geert+renesas@glider.be> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20220601070707.3946847-5-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-06-01 07:07:00 +00:00
} else if (dev_is_best_effort(dev) &&
link->flags & DL_FLAG_INFERRED &&
link->status != DL_STATE_CONSUMER_PROBE &&
!link->supplier->can_match) {
/*
* When dev_is_best_effort() is true, we ignore device
* links to suppliers that don't have a driver. If the
* consumer device still managed to probe, there's no
* point in maintaining a device link in a weird state
* (consumer probed before supplier). So delete it.
*/
device_link_drop_managed(link);
driver core: Fix SYNC_STATE_ONLY device link implementation When SYNC_STATE_ONLY support was added in commit 05ef983e0d65 ("driver core: Add device link support for SYNC_STATE_ONLY flag"), device_link_add() incorrectly skipped adding the new SYNC_STATE_ONLY device link to the supplier's and consumer's "device link" list. This causes multiple issues: - The device link is lost forever from driver core if the caller didn't keep track of it (caller typically isn't expected to). This is a memory leak. - The device link is also never visible to any other code path after device_link_add() returns. If we fix the "device link" list handling, that exposes a bunch of issues. 1. The device link "status" state management code rightfully doesn't handle the case where a DL_FLAG_MANAGED device link exists between a supplier and consumer, but the consumer manages to probe successfully before the supplier. The addition of DL_FLAG_SYNC_STATE_ONLY links break this assumption. This causes device_links_driver_bound() to throw a warning when this happens. Since DL_FLAG_SYNC_STATE_ONLY device links are mainly used for creating proxy device links for child device dependencies and aren't useful once the consumer device probes successfully, this patch just deletes DL_FLAG_SYNC_STATE_ONLY device links once its consumer device probes. This way, we avoid the warning, free up some memory and avoid complicating the device links "status" state management code. 2. Creating a DL_FLAG_STATELESS device link between two devices that already have a DL_FLAG_SYNC_STATE_ONLY device link will result in the DL_FLAG_STATELESS flag not getting set correctly. This patch also fixes this. Lastly, this patch also fixes minor whitespace issues. Cc: stable@vger.kernel.org Fixes: 05ef983e0d65 ("driver core: Add device link support for SYNC_STATE_ONLY flag") Signed-off-by: Saravana Kannan <saravanak@google.com> Reviewed-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Link: https://lore.kernel.org/r/20200519063000.128819-1-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-05-19 06:30:00 +00:00
} else {
WARN_ON(link->status != DL_STATE_CONSUMER_PROBE);
WRITE_ONCE(link->status, DL_STATE_ACTIVE);
}
driver core: Add sync_state driver/bus callback This sync_state driver/bus callback is called once all the consumers of a supplier have probed successfully. This allows the supplier device's driver/bus to sync the supplier device's state to the software state with the guarantee that all the consumers are actively managing the resources provided by the supplier device. To maintain backwards compatibility and ease transition from existing frameworks and resource cleanup schemes, late_initcall_sync is the earliest when the sync_state callback might be called. There is no upper bound on the time by which the sync_state callback has to be called. This is because if a consumer device never probes, the supplier has to maintain its resources in the state left by the bootloader. For example, if the bootloader leaves the display backlight at a fixed voltage and the backlight driver is never probed, you don't want the backlight to ever be turned off after boot up. Also, when multiple devices are added after kernel init, some suppliers could be added before their consumer devices get added. In these instances, the supplier devices could get their sync_state callback called right after they probe because the consumers devices haven't had a chance to create device links to the suppliers. To handle this correctly, this change also provides APIs to pause/resume sync state callbacks so that when multiple devices are added, their sync_state callback evaluation can be postponed to happen after all of them are added. kbuild test robot reported missing documentation for device.state_synced Reported-by: kbuild test robot <lkp@intel.com> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20190904211126.47518-5-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-09-04 21:11:23 +00:00
driver core: Fix SYNC_STATE_ONLY device link implementation When SYNC_STATE_ONLY support was added in commit 05ef983e0d65 ("driver core: Add device link support for SYNC_STATE_ONLY flag"), device_link_add() incorrectly skipped adding the new SYNC_STATE_ONLY device link to the supplier's and consumer's "device link" list. This causes multiple issues: - The device link is lost forever from driver core if the caller didn't keep track of it (caller typically isn't expected to). This is a memory leak. - The device link is also never visible to any other code path after device_link_add() returns. If we fix the "device link" list handling, that exposes a bunch of issues. 1. The device link "status" state management code rightfully doesn't handle the case where a DL_FLAG_MANAGED device link exists between a supplier and consumer, but the consumer manages to probe successfully before the supplier. The addition of DL_FLAG_SYNC_STATE_ONLY links break this assumption. This causes device_links_driver_bound() to throw a warning when this happens. Since DL_FLAG_SYNC_STATE_ONLY device links are mainly used for creating proxy device links for child device dependencies and aren't useful once the consumer device probes successfully, this patch just deletes DL_FLAG_SYNC_STATE_ONLY device links once its consumer device probes. This way, we avoid the warning, free up some memory and avoid complicating the device links "status" state management code. 2. Creating a DL_FLAG_STATELESS device link between two devices that already have a DL_FLAG_SYNC_STATE_ONLY device link will result in the DL_FLAG_STATELESS flag not getting set correctly. This patch also fixes this. Lastly, this patch also fixes minor whitespace issues. Cc: stable@vger.kernel.org Fixes: 05ef983e0d65 ("driver core: Add device link support for SYNC_STATE_ONLY flag") Signed-off-by: Saravana Kannan <saravanak@google.com> Reviewed-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Link: https://lore.kernel.org/r/20200519063000.128819-1-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-05-19 06:30:00 +00:00
/*
* This needs to be done even for the deleted
* DL_FLAG_SYNC_STATE_ONLY device link in case it was the last
* device link that was preventing the supplier from getting a
* sync_state() call.
*/
driver core: Add sync_state driver/bus callback This sync_state driver/bus callback is called once all the consumers of a supplier have probed successfully. This allows the supplier device's driver/bus to sync the supplier device's state to the software state with the guarantee that all the consumers are actively managing the resources provided by the supplier device. To maintain backwards compatibility and ease transition from existing frameworks and resource cleanup schemes, late_initcall_sync is the earliest when the sync_state callback might be called. There is no upper bound on the time by which the sync_state callback has to be called. This is because if a consumer device never probes, the supplier has to maintain its resources in the state left by the bootloader. For example, if the bootloader leaves the display backlight at a fixed voltage and the backlight driver is never probed, you don't want the backlight to ever be turned off after boot up. Also, when multiple devices are added after kernel init, some suppliers could be added before their consumer devices get added. In these instances, the supplier devices could get their sync_state callback called right after they probe because the consumers devices haven't had a chance to create device links to the suppliers. To handle this correctly, this change also provides APIs to pause/resume sync state callbacks so that when multiple devices are added, their sync_state callback evaluation can be postponed to happen after all of them are added. kbuild test robot reported missing documentation for device.state_synced Reported-by: kbuild test robot <lkp@intel.com> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20190904211126.47518-5-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-09-04 21:11:23 +00:00
if (defer_sync_state_count)
driver core: Fix SYNC_STATE_ONLY device link implementation When SYNC_STATE_ONLY support was added in commit 05ef983e0d65 ("driver core: Add device link support for SYNC_STATE_ONLY flag"), device_link_add() incorrectly skipped adding the new SYNC_STATE_ONLY device link to the supplier's and consumer's "device link" list. This causes multiple issues: - The device link is lost forever from driver core if the caller didn't keep track of it (caller typically isn't expected to). This is a memory leak. - The device link is also never visible to any other code path after device_link_add() returns. If we fix the "device link" list handling, that exposes a bunch of issues. 1. The device link "status" state management code rightfully doesn't handle the case where a DL_FLAG_MANAGED device link exists between a supplier and consumer, but the consumer manages to probe successfully before the supplier. The addition of DL_FLAG_SYNC_STATE_ONLY links break this assumption. This causes device_links_driver_bound() to throw a warning when this happens. Since DL_FLAG_SYNC_STATE_ONLY device links are mainly used for creating proxy device links for child device dependencies and aren't useful once the consumer device probes successfully, this patch just deletes DL_FLAG_SYNC_STATE_ONLY device links once its consumer device probes. This way, we avoid the warning, free up some memory and avoid complicating the device links "status" state management code. 2. Creating a DL_FLAG_STATELESS device link between two devices that already have a DL_FLAG_SYNC_STATE_ONLY device link will result in the DL_FLAG_STATELESS flag not getting set correctly. This patch also fixes this. Lastly, this patch also fixes minor whitespace issues. Cc: stable@vger.kernel.org Fixes: 05ef983e0d65 ("driver core: Add device link support for SYNC_STATE_ONLY flag") Signed-off-by: Saravana Kannan <saravanak@google.com> Reviewed-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Link: https://lore.kernel.org/r/20200519063000.128819-1-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-05-19 06:30:00 +00:00
__device_links_supplier_defer_sync(supplier);
driver core: Add sync_state driver/bus callback This sync_state driver/bus callback is called once all the consumers of a supplier have probed successfully. This allows the supplier device's driver/bus to sync the supplier device's state to the software state with the guarantee that all the consumers are actively managing the resources provided by the supplier device. To maintain backwards compatibility and ease transition from existing frameworks and resource cleanup schemes, late_initcall_sync is the earliest when the sync_state callback might be called. There is no upper bound on the time by which the sync_state callback has to be called. This is because if a consumer device never probes, the supplier has to maintain its resources in the state left by the bootloader. For example, if the bootloader leaves the display backlight at a fixed voltage and the backlight driver is never probed, you don't want the backlight to ever be turned off after boot up. Also, when multiple devices are added after kernel init, some suppliers could be added before their consumer devices get added. In these instances, the supplier devices could get their sync_state callback called right after they probe because the consumers devices haven't had a chance to create device links to the suppliers. To handle this correctly, this change also provides APIs to pause/resume sync state callbacks so that when multiple devices are added, their sync_state callback evaluation can be postponed to happen after all of them are added. kbuild test robot reported missing documentation for device.state_synced Reported-by: kbuild test robot <lkp@intel.com> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20190904211126.47518-5-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-09-04 21:11:23 +00:00
else
driver core: Fix SYNC_STATE_ONLY device link implementation When SYNC_STATE_ONLY support was added in commit 05ef983e0d65 ("driver core: Add device link support for SYNC_STATE_ONLY flag"), device_link_add() incorrectly skipped adding the new SYNC_STATE_ONLY device link to the supplier's and consumer's "device link" list. This causes multiple issues: - The device link is lost forever from driver core if the caller didn't keep track of it (caller typically isn't expected to). This is a memory leak. - The device link is also never visible to any other code path after device_link_add() returns. If we fix the "device link" list handling, that exposes a bunch of issues. 1. The device link "status" state management code rightfully doesn't handle the case where a DL_FLAG_MANAGED device link exists between a supplier and consumer, but the consumer manages to probe successfully before the supplier. The addition of DL_FLAG_SYNC_STATE_ONLY links break this assumption. This causes device_links_driver_bound() to throw a warning when this happens. Since DL_FLAG_SYNC_STATE_ONLY device links are mainly used for creating proxy device links for child device dependencies and aren't useful once the consumer device probes successfully, this patch just deletes DL_FLAG_SYNC_STATE_ONLY device links once its consumer device probes. This way, we avoid the warning, free up some memory and avoid complicating the device links "status" state management code. 2. Creating a DL_FLAG_STATELESS device link between two devices that already have a DL_FLAG_SYNC_STATE_ONLY device link will result in the DL_FLAG_STATELESS flag not getting set correctly. This patch also fixes this. Lastly, this patch also fixes minor whitespace issues. Cc: stable@vger.kernel.org Fixes: 05ef983e0d65 ("driver core: Add device link support for SYNC_STATE_ONLY flag") Signed-off-by: Saravana Kannan <saravanak@google.com> Reviewed-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Link: https://lore.kernel.org/r/20200519063000.128819-1-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-05-19 06:30:00 +00:00
__device_links_queue_sync_state(supplier, &sync_list);
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
}
dev->links.status = DL_DEV_DRIVER_BOUND;
device_links_write_unlock();
device_links_flush_sync_list(&sync_list, dev);
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
}
/**
* __device_links_no_driver - Update links of a device without a driver.
* @dev: Device without a drvier.
*
* Delete all non-persistent links from this device to any suppliers.
*
* Persistent links stay around, but their status is changed to "available",
* unless they already are in the "supplier unbind in progress" state in which
* case they need not be updated.
*
driver core: Remove device link creation limitation If device_link_add() is called for a consumer/supplier pair with an existing device link between them and the existing link's type is not in agreement with the flags passed to that function by its caller, NULL will be returned. That is seriously inconvenient, because it forces the callers of device_link_add() to worry about what others may or may not do even if that is not relevant to them for any other reasons. It turns out, however, that this limitation can be made go away relatively easily. The underlying observation is that if DL_FLAG_STATELESS has been passed to device_link_add() in flags for the given consumer/supplier pair at least once, calling either device_link_del() or device_link_remove() to release the link returned by it should work, but there are no other requirements associated with that flag. In turn, if at least one of the callers of device_link_add() for the given consumer/supplier pair has not passed DL_FLAG_STATELESS to it in flags, the driver core should track the status of the link and act on it as appropriate (ie. the link should be treated as "managed"). This means that DL_FLAG_STATELESS needs to be set for managed device links and it should be valid to call device_link_del() or device_link_remove() to drop references to them in certain sutiations. To allow that to happen, introduce a new (internal) device link flag called DL_FLAG_MANAGED and make device_link_add() set it automatically whenever DL_FLAG_STATELESS is not passed to it. Also make it take additional references to existing device links that were previously stateless (that is, with DL_FLAG_STATELESS set and DL_FLAG_MANAGED unset) and will need to be managed going forward and initialize their status (which has been DL_STATE_NONE so far). Accordingly, when a managed device link is dropped automatically by the driver core, make it clear DL_FLAG_MANAGED, reset the link's status back to DL_STATE_NONE and drop the reference to it associated with DL_FLAG_MANAGED instead of just deleting it right away (to allow it to stay around in case it still needs to be released explicitly by someone). With that, since setting DL_FLAG_STATELESS doesn't mean that the device link in question is not managed any more, replace all of the status-tracking checks against DL_FLAG_STATELESS with analogous checks against DL_FLAG_MANAGED and update the documentation to reflect these changes. While at it, make device_link_add() reject flags that it does not recognize, including DL_FLAG_MANAGED. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Reviewed-by: Saravana Kannan <saravanak@google.com> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Review-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/2305283.AStDPdUUnE@kreacher Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-07-16 15:21:06 +00:00
* Links without the DL_FLAG_MANAGED flag set are ignored.
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
*/
static void __device_links_no_driver(struct device *dev)
{
struct device_link *link, *ln;
list_for_each_entry_safe_reverse(link, ln, &dev->links.suppliers, c_node) {
driver core: Remove device link creation limitation If device_link_add() is called for a consumer/supplier pair with an existing device link between them and the existing link's type is not in agreement with the flags passed to that function by its caller, NULL will be returned. That is seriously inconvenient, because it forces the callers of device_link_add() to worry about what others may or may not do even if that is not relevant to them for any other reasons. It turns out, however, that this limitation can be made go away relatively easily. The underlying observation is that if DL_FLAG_STATELESS has been passed to device_link_add() in flags for the given consumer/supplier pair at least once, calling either device_link_del() or device_link_remove() to release the link returned by it should work, but there are no other requirements associated with that flag. In turn, if at least one of the callers of device_link_add() for the given consumer/supplier pair has not passed DL_FLAG_STATELESS to it in flags, the driver core should track the status of the link and act on it as appropriate (ie. the link should be treated as "managed"). This means that DL_FLAG_STATELESS needs to be set for managed device links and it should be valid to call device_link_del() or device_link_remove() to drop references to them in certain sutiations. To allow that to happen, introduce a new (internal) device link flag called DL_FLAG_MANAGED and make device_link_add() set it automatically whenever DL_FLAG_STATELESS is not passed to it. Also make it take additional references to existing device links that were previously stateless (that is, with DL_FLAG_STATELESS set and DL_FLAG_MANAGED unset) and will need to be managed going forward and initialize their status (which has been DL_STATE_NONE so far). Accordingly, when a managed device link is dropped automatically by the driver core, make it clear DL_FLAG_MANAGED, reset the link's status back to DL_STATE_NONE and drop the reference to it associated with DL_FLAG_MANAGED instead of just deleting it right away (to allow it to stay around in case it still needs to be released explicitly by someone). With that, since setting DL_FLAG_STATELESS doesn't mean that the device link in question is not managed any more, replace all of the status-tracking checks against DL_FLAG_STATELESS with analogous checks against DL_FLAG_MANAGED and update the documentation to reflect these changes. While at it, make device_link_add() reject flags that it does not recognize, including DL_FLAG_MANAGED. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Reviewed-by: Saravana Kannan <saravanak@google.com> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Review-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/2305283.AStDPdUUnE@kreacher Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-07-16 15:21:06 +00:00
if (!(link->flags & DL_FLAG_MANAGED))
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
continue;
driver core: Update device link status correctly for SYNC_STATE_ONLY links When SYNC_STATE_ONLY support was added in commit 05ef983e0d65 ("driver core: Add device link support for SYNC_STATE_ONLY flag"), SYNC_STATE_ONLY links were treated similar to STATELESS links in terms of not blocking consumer probe if the supplier hasn't probed yet. That caused a SYNC_STATE_ONLY device link's status to not get updated. Since SYNC_STATE_ONLY device link is no longer useful once the consumer probes, commit 21c27f06587d ("driver core: Fix SYNC_STATE_ONLY device link implementation") addresses the status update issue by deleting the SYNC_STATE_ONLY device link instead of complicating the status update code. However, there are still some cases where we need to update the status of a SYNC_STATE_ONLY device link. This is because a SYNC_STATE_ONLY device link can later get converted into a normal MANAGED device link when a normal MANAGED device link is created between a supplier and consumer that already have a SYNC_STATE_ONLY device link between them. If a SYNC_STATE_ONLY device link's status isn't maintained correctly till it's converted to a normal MANAGED device link, then the normal MANAGED device link will end up with a wrong link status. This can cause a warning stack trace[1] when the consumer device probes successfully. This commit fixes the SYNC_STATE_ONLY device link status update issue where it wouldn't transition correctly from DL_STATE_DORMANT or DL_STATE_AVAILABLE to DL_STATE_CONSUMER_PROBE. It also resets the status back to DL_STATE_DORMANT or DL_STATE_AVAILABLE if the consumer probe fails. [1] - https://lore.kernel.org/lkml/20200522204120.3b3c9ed6@apollo/ Fixes: 05ef983e0d65 ("driver core: Add device link support for SYNC_STATE_ONLY flag") Fixes: 21c27f06587d ("driver core: Fix SYNC_STATE_ONLY device link implementation") Reported-by: Michael Walle <michael@walle.cc> Tested-by: Michael Walle <michael@walle.cc> Signed-off-by: Saravana Kannan <saravanak@google.com> Reviewed-by: Rafael J. Wysocki <rrafael.j.wysocki@intel.com> Link: https://lore.kernel.org/r/20200526220928.49939-1-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-05-26 22:09:27 +00:00
if (link->flags & DL_FLAG_AUTOREMOVE_CONSUMER) {
driver core: Remove device link creation limitation If device_link_add() is called for a consumer/supplier pair with an existing device link between them and the existing link's type is not in agreement with the flags passed to that function by its caller, NULL will be returned. That is seriously inconvenient, because it forces the callers of device_link_add() to worry about what others may or may not do even if that is not relevant to them for any other reasons. It turns out, however, that this limitation can be made go away relatively easily. The underlying observation is that if DL_FLAG_STATELESS has been passed to device_link_add() in flags for the given consumer/supplier pair at least once, calling either device_link_del() or device_link_remove() to release the link returned by it should work, but there are no other requirements associated with that flag. In turn, if at least one of the callers of device_link_add() for the given consumer/supplier pair has not passed DL_FLAG_STATELESS to it in flags, the driver core should track the status of the link and act on it as appropriate (ie. the link should be treated as "managed"). This means that DL_FLAG_STATELESS needs to be set for managed device links and it should be valid to call device_link_del() or device_link_remove() to drop references to them in certain sutiations. To allow that to happen, introduce a new (internal) device link flag called DL_FLAG_MANAGED and make device_link_add() set it automatically whenever DL_FLAG_STATELESS is not passed to it. Also make it take additional references to existing device links that were previously stateless (that is, with DL_FLAG_STATELESS set and DL_FLAG_MANAGED unset) and will need to be managed going forward and initialize their status (which has been DL_STATE_NONE so far). Accordingly, when a managed device link is dropped automatically by the driver core, make it clear DL_FLAG_MANAGED, reset the link's status back to DL_STATE_NONE and drop the reference to it associated with DL_FLAG_MANAGED instead of just deleting it right away (to allow it to stay around in case it still needs to be released explicitly by someone). With that, since setting DL_FLAG_STATELESS doesn't mean that the device link in question is not managed any more, replace all of the status-tracking checks against DL_FLAG_STATELESS with analogous checks against DL_FLAG_MANAGED and update the documentation to reflect these changes. While at it, make device_link_add() reject flags that it does not recognize, including DL_FLAG_MANAGED. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Reviewed-by: Saravana Kannan <saravanak@google.com> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Review-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/2305283.AStDPdUUnE@kreacher Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-07-16 15:21:06 +00:00
device_link_drop_managed(link);
driver core: Update device link status correctly for SYNC_STATE_ONLY links When SYNC_STATE_ONLY support was added in commit 05ef983e0d65 ("driver core: Add device link support for SYNC_STATE_ONLY flag"), SYNC_STATE_ONLY links were treated similar to STATELESS links in terms of not blocking consumer probe if the supplier hasn't probed yet. That caused a SYNC_STATE_ONLY device link's status to not get updated. Since SYNC_STATE_ONLY device link is no longer useful once the consumer probes, commit 21c27f06587d ("driver core: Fix SYNC_STATE_ONLY device link implementation") addresses the status update issue by deleting the SYNC_STATE_ONLY device link instead of complicating the status update code. However, there are still some cases where we need to update the status of a SYNC_STATE_ONLY device link. This is because a SYNC_STATE_ONLY device link can later get converted into a normal MANAGED device link when a normal MANAGED device link is created between a supplier and consumer that already have a SYNC_STATE_ONLY device link between them. If a SYNC_STATE_ONLY device link's status isn't maintained correctly till it's converted to a normal MANAGED device link, then the normal MANAGED device link will end up with a wrong link status. This can cause a warning stack trace[1] when the consumer device probes successfully. This commit fixes the SYNC_STATE_ONLY device link status update issue where it wouldn't transition correctly from DL_STATE_DORMANT or DL_STATE_AVAILABLE to DL_STATE_CONSUMER_PROBE. It also resets the status back to DL_STATE_DORMANT or DL_STATE_AVAILABLE if the consumer probe fails. [1] - https://lore.kernel.org/lkml/20200522204120.3b3c9ed6@apollo/ Fixes: 05ef983e0d65 ("driver core: Add device link support for SYNC_STATE_ONLY flag") Fixes: 21c27f06587d ("driver core: Fix SYNC_STATE_ONLY device link implementation") Reported-by: Michael Walle <michael@walle.cc> Tested-by: Michael Walle <michael@walle.cc> Signed-off-by: Saravana Kannan <saravanak@google.com> Reviewed-by: Rafael J. Wysocki <rrafael.j.wysocki@intel.com> Link: https://lore.kernel.org/r/20200526220928.49939-1-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-05-26 22:09:27 +00:00
continue;
}
if (link->status != DL_STATE_CONSUMER_PROBE &&
link->status != DL_STATE_ACTIVE)
continue;
if (link->supplier->links.status == DL_DEV_DRIVER_BOUND) {
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
WRITE_ONCE(link->status, DL_STATE_AVAILABLE);
driver core: Update device link status correctly for SYNC_STATE_ONLY links When SYNC_STATE_ONLY support was added in commit 05ef983e0d65 ("driver core: Add device link support for SYNC_STATE_ONLY flag"), SYNC_STATE_ONLY links were treated similar to STATELESS links in terms of not blocking consumer probe if the supplier hasn't probed yet. That caused a SYNC_STATE_ONLY device link's status to not get updated. Since SYNC_STATE_ONLY device link is no longer useful once the consumer probes, commit 21c27f06587d ("driver core: Fix SYNC_STATE_ONLY device link implementation") addresses the status update issue by deleting the SYNC_STATE_ONLY device link instead of complicating the status update code. However, there are still some cases where we need to update the status of a SYNC_STATE_ONLY device link. This is because a SYNC_STATE_ONLY device link can later get converted into a normal MANAGED device link when a normal MANAGED device link is created between a supplier and consumer that already have a SYNC_STATE_ONLY device link between them. If a SYNC_STATE_ONLY device link's status isn't maintained correctly till it's converted to a normal MANAGED device link, then the normal MANAGED device link will end up with a wrong link status. This can cause a warning stack trace[1] when the consumer device probes successfully. This commit fixes the SYNC_STATE_ONLY device link status update issue where it wouldn't transition correctly from DL_STATE_DORMANT or DL_STATE_AVAILABLE to DL_STATE_CONSUMER_PROBE. It also resets the status back to DL_STATE_DORMANT or DL_STATE_AVAILABLE if the consumer probe fails. [1] - https://lore.kernel.org/lkml/20200522204120.3b3c9ed6@apollo/ Fixes: 05ef983e0d65 ("driver core: Add device link support for SYNC_STATE_ONLY flag") Fixes: 21c27f06587d ("driver core: Fix SYNC_STATE_ONLY device link implementation") Reported-by: Michael Walle <michael@walle.cc> Tested-by: Michael Walle <michael@walle.cc> Signed-off-by: Saravana Kannan <saravanak@google.com> Reviewed-by: Rafael J. Wysocki <rrafael.j.wysocki@intel.com> Link: https://lore.kernel.org/r/20200526220928.49939-1-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-05-26 22:09:27 +00:00
} else {
WARN_ON(!(link->flags & DL_FLAG_SYNC_STATE_ONLY));
WRITE_ONCE(link->status, DL_STATE_DORMANT);
}
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
}
dev->links.status = DL_DEV_NO_DRIVER;
}
driver core: Fix adding device links to probing suppliers Currently, it is not valid to add a device link from a consumer driver ->probe callback to a supplier that is still probing too, but generally this is a valid use case. For example, if the consumer has just acquired a resource that can only be available if the supplier is functional, adding a device link to that supplier right away should be safe (and even desirable arguably), but device_link_add() doesn't handle that case correctly and the initial state of the link created by it is wrong then. To address this problem, change the initial state of device links added between a probing supplier and a probing consumer to DL_STATE_CONSUMER_PROBE and update device_links_driver_bound() to skip such links on the supplier side. With this change, if the supplier probe completes first, device_links_driver_bound() called for it will skip the link state update and when it is called for the consumer, the link state will be updated to "active". In turn, if the consumer probe completes first, device_links_driver_bound() called for it will change the state of the link to "active" and when it is called for the supplier, the link status update will be skipped. However, in principle the supplier or consumer probe may still fail after the link has been added, so modify device_links_no_driver() to change device links in the "active" or "consumer probe" state to "dormant" on the supplier side and update __device_links_no_driver() to change the link state to "available" only if it is "consumer probe" or "active". Then, if the supplier probe fails first, the leftover link to the probing consumer will become "dormant" and device_links_no_driver() called for the consumer (when its probe fails) will clean it up. In turn, if the consumer probe fails first, it will either drop the link, or change its state to "available" and, in the latter case, when device_links_no_driver() is called for the supplier, it will update the link state to "dormant". [If the supplier probe fails, but the consumer probe succeeds, which should not happen as long as the consumer driver is correct, the link still will be around, but it will be "dormant" until the supplier is probed again.] Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-02-01 00:50:39 +00:00
/**
* device_links_no_driver - Update links after failing driver probe.
* @dev: Device whose driver has just failed to probe.
*
* Clean up leftover links to consumers for @dev and invoke
* %__device_links_no_driver() to update links to suppliers for it as
* appropriate.
*
driver core: Remove device link creation limitation If device_link_add() is called for a consumer/supplier pair with an existing device link between them and the existing link's type is not in agreement with the flags passed to that function by its caller, NULL will be returned. That is seriously inconvenient, because it forces the callers of device_link_add() to worry about what others may or may not do even if that is not relevant to them for any other reasons. It turns out, however, that this limitation can be made go away relatively easily. The underlying observation is that if DL_FLAG_STATELESS has been passed to device_link_add() in flags for the given consumer/supplier pair at least once, calling either device_link_del() or device_link_remove() to release the link returned by it should work, but there are no other requirements associated with that flag. In turn, if at least one of the callers of device_link_add() for the given consumer/supplier pair has not passed DL_FLAG_STATELESS to it in flags, the driver core should track the status of the link and act on it as appropriate (ie. the link should be treated as "managed"). This means that DL_FLAG_STATELESS needs to be set for managed device links and it should be valid to call device_link_del() or device_link_remove() to drop references to them in certain sutiations. To allow that to happen, introduce a new (internal) device link flag called DL_FLAG_MANAGED and make device_link_add() set it automatically whenever DL_FLAG_STATELESS is not passed to it. Also make it take additional references to existing device links that were previously stateless (that is, with DL_FLAG_STATELESS set and DL_FLAG_MANAGED unset) and will need to be managed going forward and initialize their status (which has been DL_STATE_NONE so far). Accordingly, when a managed device link is dropped automatically by the driver core, make it clear DL_FLAG_MANAGED, reset the link's status back to DL_STATE_NONE and drop the reference to it associated with DL_FLAG_MANAGED instead of just deleting it right away (to allow it to stay around in case it still needs to be released explicitly by someone). With that, since setting DL_FLAG_STATELESS doesn't mean that the device link in question is not managed any more, replace all of the status-tracking checks against DL_FLAG_STATELESS with analogous checks against DL_FLAG_MANAGED and update the documentation to reflect these changes. While at it, make device_link_add() reject flags that it does not recognize, including DL_FLAG_MANAGED. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Reviewed-by: Saravana Kannan <saravanak@google.com> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Review-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/2305283.AStDPdUUnE@kreacher Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-07-16 15:21:06 +00:00
* Links without the DL_FLAG_MANAGED flag set are ignored.
driver core: Fix adding device links to probing suppliers Currently, it is not valid to add a device link from a consumer driver ->probe callback to a supplier that is still probing too, but generally this is a valid use case. For example, if the consumer has just acquired a resource that can only be available if the supplier is functional, adding a device link to that supplier right away should be safe (and even desirable arguably), but device_link_add() doesn't handle that case correctly and the initial state of the link created by it is wrong then. To address this problem, change the initial state of device links added between a probing supplier and a probing consumer to DL_STATE_CONSUMER_PROBE and update device_links_driver_bound() to skip such links on the supplier side. With this change, if the supplier probe completes first, device_links_driver_bound() called for it will skip the link state update and when it is called for the consumer, the link state will be updated to "active". In turn, if the consumer probe completes first, device_links_driver_bound() called for it will change the state of the link to "active" and when it is called for the supplier, the link status update will be skipped. However, in principle the supplier or consumer probe may still fail after the link has been added, so modify device_links_no_driver() to change device links in the "active" or "consumer probe" state to "dormant" on the supplier side and update __device_links_no_driver() to change the link state to "available" only if it is "consumer probe" or "active". Then, if the supplier probe fails first, the leftover link to the probing consumer will become "dormant" and device_links_no_driver() called for the consumer (when its probe fails) will clean it up. In turn, if the consumer probe fails first, it will either drop the link, or change its state to "available" and, in the latter case, when device_links_no_driver() is called for the supplier, it will update the link state to "dormant". [If the supplier probe fails, but the consumer probe succeeds, which should not happen as long as the consumer driver is correct, the link still will be around, but it will be "dormant" until the supplier is probed again.] Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-02-01 00:50:39 +00:00
*/
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
void device_links_no_driver(struct device *dev)
{
driver core: Fix adding device links to probing suppliers Currently, it is not valid to add a device link from a consumer driver ->probe callback to a supplier that is still probing too, but generally this is a valid use case. For example, if the consumer has just acquired a resource that can only be available if the supplier is functional, adding a device link to that supplier right away should be safe (and even desirable arguably), but device_link_add() doesn't handle that case correctly and the initial state of the link created by it is wrong then. To address this problem, change the initial state of device links added between a probing supplier and a probing consumer to DL_STATE_CONSUMER_PROBE and update device_links_driver_bound() to skip such links on the supplier side. With this change, if the supplier probe completes first, device_links_driver_bound() called for it will skip the link state update and when it is called for the consumer, the link state will be updated to "active". In turn, if the consumer probe completes first, device_links_driver_bound() called for it will change the state of the link to "active" and when it is called for the supplier, the link status update will be skipped. However, in principle the supplier or consumer probe may still fail after the link has been added, so modify device_links_no_driver() to change device links in the "active" or "consumer probe" state to "dormant" on the supplier side and update __device_links_no_driver() to change the link state to "available" only if it is "consumer probe" or "active". Then, if the supplier probe fails first, the leftover link to the probing consumer will become "dormant" and device_links_no_driver() called for the consumer (when its probe fails) will clean it up. In turn, if the consumer probe fails first, it will either drop the link, or change its state to "available" and, in the latter case, when device_links_no_driver() is called for the supplier, it will update the link state to "dormant". [If the supplier probe fails, but the consumer probe succeeds, which should not happen as long as the consumer driver is correct, the link still will be around, but it will be "dormant" until the supplier is probed again.] Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-02-01 00:50:39 +00:00
struct device_link *link;
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
device_links_write_lock();
driver core: Fix adding device links to probing suppliers Currently, it is not valid to add a device link from a consumer driver ->probe callback to a supplier that is still probing too, but generally this is a valid use case. For example, if the consumer has just acquired a resource that can only be available if the supplier is functional, adding a device link to that supplier right away should be safe (and even desirable arguably), but device_link_add() doesn't handle that case correctly and the initial state of the link created by it is wrong then. To address this problem, change the initial state of device links added between a probing supplier and a probing consumer to DL_STATE_CONSUMER_PROBE and update device_links_driver_bound() to skip such links on the supplier side. With this change, if the supplier probe completes first, device_links_driver_bound() called for it will skip the link state update and when it is called for the consumer, the link state will be updated to "active". In turn, if the consumer probe completes first, device_links_driver_bound() called for it will change the state of the link to "active" and when it is called for the supplier, the link status update will be skipped. However, in principle the supplier or consumer probe may still fail after the link has been added, so modify device_links_no_driver() to change device links in the "active" or "consumer probe" state to "dormant" on the supplier side and update __device_links_no_driver() to change the link state to "available" only if it is "consumer probe" or "active". Then, if the supplier probe fails first, the leftover link to the probing consumer will become "dormant" and device_links_no_driver() called for the consumer (when its probe fails) will clean it up. In turn, if the consumer probe fails first, it will either drop the link, or change its state to "available" and, in the latter case, when device_links_no_driver() is called for the supplier, it will update the link state to "dormant". [If the supplier probe fails, but the consumer probe succeeds, which should not happen as long as the consumer driver is correct, the link still will be around, but it will be "dormant" until the supplier is probed again.] Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-02-01 00:50:39 +00:00
list_for_each_entry(link, &dev->links.consumers, s_node) {
driver core: Remove device link creation limitation If device_link_add() is called for a consumer/supplier pair with an existing device link between them and the existing link's type is not in agreement with the flags passed to that function by its caller, NULL will be returned. That is seriously inconvenient, because it forces the callers of device_link_add() to worry about what others may or may not do even if that is not relevant to them for any other reasons. It turns out, however, that this limitation can be made go away relatively easily. The underlying observation is that if DL_FLAG_STATELESS has been passed to device_link_add() in flags for the given consumer/supplier pair at least once, calling either device_link_del() or device_link_remove() to release the link returned by it should work, but there are no other requirements associated with that flag. In turn, if at least one of the callers of device_link_add() for the given consumer/supplier pair has not passed DL_FLAG_STATELESS to it in flags, the driver core should track the status of the link and act on it as appropriate (ie. the link should be treated as "managed"). This means that DL_FLAG_STATELESS needs to be set for managed device links and it should be valid to call device_link_del() or device_link_remove() to drop references to them in certain sutiations. To allow that to happen, introduce a new (internal) device link flag called DL_FLAG_MANAGED and make device_link_add() set it automatically whenever DL_FLAG_STATELESS is not passed to it. Also make it take additional references to existing device links that were previously stateless (that is, with DL_FLAG_STATELESS set and DL_FLAG_MANAGED unset) and will need to be managed going forward and initialize their status (which has been DL_STATE_NONE so far). Accordingly, when a managed device link is dropped automatically by the driver core, make it clear DL_FLAG_MANAGED, reset the link's status back to DL_STATE_NONE and drop the reference to it associated with DL_FLAG_MANAGED instead of just deleting it right away (to allow it to stay around in case it still needs to be released explicitly by someone). With that, since setting DL_FLAG_STATELESS doesn't mean that the device link in question is not managed any more, replace all of the status-tracking checks against DL_FLAG_STATELESS with analogous checks against DL_FLAG_MANAGED and update the documentation to reflect these changes. While at it, make device_link_add() reject flags that it does not recognize, including DL_FLAG_MANAGED. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Reviewed-by: Saravana Kannan <saravanak@google.com> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Review-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/2305283.AStDPdUUnE@kreacher Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-07-16 15:21:06 +00:00
if (!(link->flags & DL_FLAG_MANAGED))
driver core: Fix adding device links to probing suppliers Currently, it is not valid to add a device link from a consumer driver ->probe callback to a supplier that is still probing too, but generally this is a valid use case. For example, if the consumer has just acquired a resource that can only be available if the supplier is functional, adding a device link to that supplier right away should be safe (and even desirable arguably), but device_link_add() doesn't handle that case correctly and the initial state of the link created by it is wrong then. To address this problem, change the initial state of device links added between a probing supplier and a probing consumer to DL_STATE_CONSUMER_PROBE and update device_links_driver_bound() to skip such links on the supplier side. With this change, if the supplier probe completes first, device_links_driver_bound() called for it will skip the link state update and when it is called for the consumer, the link state will be updated to "active". In turn, if the consumer probe completes first, device_links_driver_bound() called for it will change the state of the link to "active" and when it is called for the supplier, the link status update will be skipped. However, in principle the supplier or consumer probe may still fail after the link has been added, so modify device_links_no_driver() to change device links in the "active" or "consumer probe" state to "dormant" on the supplier side and update __device_links_no_driver() to change the link state to "available" only if it is "consumer probe" or "active". Then, if the supplier probe fails first, the leftover link to the probing consumer will become "dormant" and device_links_no_driver() called for the consumer (when its probe fails) will clean it up. In turn, if the consumer probe fails first, it will either drop the link, or change its state to "available" and, in the latter case, when device_links_no_driver() is called for the supplier, it will update the link state to "dormant". [If the supplier probe fails, but the consumer probe succeeds, which should not happen as long as the consumer driver is correct, the link still will be around, but it will be "dormant" until the supplier is probed again.] Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-02-01 00:50:39 +00:00
continue;
/*
* The probe has failed, so if the status of the link is
* "consumer probe" or "active", it must have been added by
* a probing consumer while this device was still probing.
* Change its state to "dormant", as it represents a valid
* relationship, but it is not functionally meaningful.
*/
if (link->status == DL_STATE_CONSUMER_PROBE ||
link->status == DL_STATE_ACTIVE)
WRITE_ONCE(link->status, DL_STATE_DORMANT);
}
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
__device_links_no_driver(dev);
driver core: Fix adding device links to probing suppliers Currently, it is not valid to add a device link from a consumer driver ->probe callback to a supplier that is still probing too, but generally this is a valid use case. For example, if the consumer has just acquired a resource that can only be available if the supplier is functional, adding a device link to that supplier right away should be safe (and even desirable arguably), but device_link_add() doesn't handle that case correctly and the initial state of the link created by it is wrong then. To address this problem, change the initial state of device links added between a probing supplier and a probing consumer to DL_STATE_CONSUMER_PROBE and update device_links_driver_bound() to skip such links on the supplier side. With this change, if the supplier probe completes first, device_links_driver_bound() called for it will skip the link state update and when it is called for the consumer, the link state will be updated to "active". In turn, if the consumer probe completes first, device_links_driver_bound() called for it will change the state of the link to "active" and when it is called for the supplier, the link status update will be skipped. However, in principle the supplier or consumer probe may still fail after the link has been added, so modify device_links_no_driver() to change device links in the "active" or "consumer probe" state to "dormant" on the supplier side and update __device_links_no_driver() to change the link state to "available" only if it is "consumer probe" or "active". Then, if the supplier probe fails first, the leftover link to the probing consumer will become "dormant" and device_links_no_driver() called for the consumer (when its probe fails) will clean it up. In turn, if the consumer probe fails first, it will either drop the link, or change its state to "available" and, in the latter case, when device_links_no_driver() is called for the supplier, it will update the link state to "dormant". [If the supplier probe fails, but the consumer probe succeeds, which should not happen as long as the consumer driver is correct, the link still will be around, but it will be "dormant" until the supplier is probed again.] Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-02-01 00:50:39 +00:00
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
device_links_write_unlock();
}
/**
* device_links_driver_cleanup - Update links after driver removal.
* @dev: Device whose driver has just gone away.
*
* Update links to consumers for @dev by changing their status to "dormant" and
* invoke %__device_links_no_driver() to update links to suppliers for it as
* appropriate.
*
driver core: Remove device link creation limitation If device_link_add() is called for a consumer/supplier pair with an existing device link between them and the existing link's type is not in agreement with the flags passed to that function by its caller, NULL will be returned. That is seriously inconvenient, because it forces the callers of device_link_add() to worry about what others may or may not do even if that is not relevant to them for any other reasons. It turns out, however, that this limitation can be made go away relatively easily. The underlying observation is that if DL_FLAG_STATELESS has been passed to device_link_add() in flags for the given consumer/supplier pair at least once, calling either device_link_del() or device_link_remove() to release the link returned by it should work, but there are no other requirements associated with that flag. In turn, if at least one of the callers of device_link_add() for the given consumer/supplier pair has not passed DL_FLAG_STATELESS to it in flags, the driver core should track the status of the link and act on it as appropriate (ie. the link should be treated as "managed"). This means that DL_FLAG_STATELESS needs to be set for managed device links and it should be valid to call device_link_del() or device_link_remove() to drop references to them in certain sutiations. To allow that to happen, introduce a new (internal) device link flag called DL_FLAG_MANAGED and make device_link_add() set it automatically whenever DL_FLAG_STATELESS is not passed to it. Also make it take additional references to existing device links that were previously stateless (that is, with DL_FLAG_STATELESS set and DL_FLAG_MANAGED unset) and will need to be managed going forward and initialize their status (which has been DL_STATE_NONE so far). Accordingly, when a managed device link is dropped automatically by the driver core, make it clear DL_FLAG_MANAGED, reset the link's status back to DL_STATE_NONE and drop the reference to it associated with DL_FLAG_MANAGED instead of just deleting it right away (to allow it to stay around in case it still needs to be released explicitly by someone). With that, since setting DL_FLAG_STATELESS doesn't mean that the device link in question is not managed any more, replace all of the status-tracking checks against DL_FLAG_STATELESS with analogous checks against DL_FLAG_MANAGED and update the documentation to reflect these changes. While at it, make device_link_add() reject flags that it does not recognize, including DL_FLAG_MANAGED. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Reviewed-by: Saravana Kannan <saravanak@google.com> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Review-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/2305283.AStDPdUUnE@kreacher Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-07-16 15:21:06 +00:00
* Links without the DL_FLAG_MANAGED flag set are ignored.
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
*/
void device_links_driver_cleanup(struct device *dev)
{
struct device_link *link, *ln;
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
device_links_write_lock();
list_for_each_entry_safe(link, ln, &dev->links.consumers, s_node) {
driver core: Remove device link creation limitation If device_link_add() is called for a consumer/supplier pair with an existing device link between them and the existing link's type is not in agreement with the flags passed to that function by its caller, NULL will be returned. That is seriously inconvenient, because it forces the callers of device_link_add() to worry about what others may or may not do even if that is not relevant to them for any other reasons. It turns out, however, that this limitation can be made go away relatively easily. The underlying observation is that if DL_FLAG_STATELESS has been passed to device_link_add() in flags for the given consumer/supplier pair at least once, calling either device_link_del() or device_link_remove() to release the link returned by it should work, but there are no other requirements associated with that flag. In turn, if at least one of the callers of device_link_add() for the given consumer/supplier pair has not passed DL_FLAG_STATELESS to it in flags, the driver core should track the status of the link and act on it as appropriate (ie. the link should be treated as "managed"). This means that DL_FLAG_STATELESS needs to be set for managed device links and it should be valid to call device_link_del() or device_link_remove() to drop references to them in certain sutiations. To allow that to happen, introduce a new (internal) device link flag called DL_FLAG_MANAGED and make device_link_add() set it automatically whenever DL_FLAG_STATELESS is not passed to it. Also make it take additional references to existing device links that were previously stateless (that is, with DL_FLAG_STATELESS set and DL_FLAG_MANAGED unset) and will need to be managed going forward and initialize their status (which has been DL_STATE_NONE so far). Accordingly, when a managed device link is dropped automatically by the driver core, make it clear DL_FLAG_MANAGED, reset the link's status back to DL_STATE_NONE and drop the reference to it associated with DL_FLAG_MANAGED instead of just deleting it right away (to allow it to stay around in case it still needs to be released explicitly by someone). With that, since setting DL_FLAG_STATELESS doesn't mean that the device link in question is not managed any more, replace all of the status-tracking checks against DL_FLAG_STATELESS with analogous checks against DL_FLAG_MANAGED and update the documentation to reflect these changes. While at it, make device_link_add() reject flags that it does not recognize, including DL_FLAG_MANAGED. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Reviewed-by: Saravana Kannan <saravanak@google.com> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Review-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/2305283.AStDPdUUnE@kreacher Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-07-16 15:21:06 +00:00
if (!(link->flags & DL_FLAG_MANAGED))
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
continue;
WARN_ON(link->flags & DL_FLAG_AUTOREMOVE_CONSUMER);
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
WARN_ON(link->status != DL_STATE_SUPPLIER_UNBIND);
/*
* autoremove the links between this @dev and its consumer
* devices that are not active, i.e. where the link state
* has moved to DL_STATE_SUPPLIER_UNBIND.
*/
if (link->status == DL_STATE_SUPPLIER_UNBIND &&
link->flags & DL_FLAG_AUTOREMOVE_SUPPLIER)
driver core: Remove device link creation limitation If device_link_add() is called for a consumer/supplier pair with an existing device link between them and the existing link's type is not in agreement with the flags passed to that function by its caller, NULL will be returned. That is seriously inconvenient, because it forces the callers of device_link_add() to worry about what others may or may not do even if that is not relevant to them for any other reasons. It turns out, however, that this limitation can be made go away relatively easily. The underlying observation is that if DL_FLAG_STATELESS has been passed to device_link_add() in flags for the given consumer/supplier pair at least once, calling either device_link_del() or device_link_remove() to release the link returned by it should work, but there are no other requirements associated with that flag. In turn, if at least one of the callers of device_link_add() for the given consumer/supplier pair has not passed DL_FLAG_STATELESS to it in flags, the driver core should track the status of the link and act on it as appropriate (ie. the link should be treated as "managed"). This means that DL_FLAG_STATELESS needs to be set for managed device links and it should be valid to call device_link_del() or device_link_remove() to drop references to them in certain sutiations. To allow that to happen, introduce a new (internal) device link flag called DL_FLAG_MANAGED and make device_link_add() set it automatically whenever DL_FLAG_STATELESS is not passed to it. Also make it take additional references to existing device links that were previously stateless (that is, with DL_FLAG_STATELESS set and DL_FLAG_MANAGED unset) and will need to be managed going forward and initialize their status (which has been DL_STATE_NONE so far). Accordingly, when a managed device link is dropped automatically by the driver core, make it clear DL_FLAG_MANAGED, reset the link's status back to DL_STATE_NONE and drop the reference to it associated with DL_FLAG_MANAGED instead of just deleting it right away (to allow it to stay around in case it still needs to be released explicitly by someone). With that, since setting DL_FLAG_STATELESS doesn't mean that the device link in question is not managed any more, replace all of the status-tracking checks against DL_FLAG_STATELESS with analogous checks against DL_FLAG_MANAGED and update the documentation to reflect these changes. While at it, make device_link_add() reject flags that it does not recognize, including DL_FLAG_MANAGED. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Reviewed-by: Saravana Kannan <saravanak@google.com> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Review-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/2305283.AStDPdUUnE@kreacher Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-07-16 15:21:06 +00:00
device_link_drop_managed(link);
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
WRITE_ONCE(link->status, DL_STATE_DORMANT);
}
list_del_init(&dev->links.defer_sync);
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
__device_links_no_driver(dev);
device_links_write_unlock();
}
/**
* device_links_busy - Check if there are any busy links to consumers.
* @dev: Device to check.
*
* Check each consumer of the device and return 'true' if its link's status
* is one of "consumer probe" or "active" (meaning that the given consumer is
* probing right now or its driver is present). Otherwise, change the link
* state to "supplier unbind" to prevent the consumer from being probed
* successfully going forward.
*
* Return 'false' if there are no probing or active consumers.
*
driver core: Remove device link creation limitation If device_link_add() is called for a consumer/supplier pair with an existing device link between them and the existing link's type is not in agreement with the flags passed to that function by its caller, NULL will be returned. That is seriously inconvenient, because it forces the callers of device_link_add() to worry about what others may or may not do even if that is not relevant to them for any other reasons. It turns out, however, that this limitation can be made go away relatively easily. The underlying observation is that if DL_FLAG_STATELESS has been passed to device_link_add() in flags for the given consumer/supplier pair at least once, calling either device_link_del() or device_link_remove() to release the link returned by it should work, but there are no other requirements associated with that flag. In turn, if at least one of the callers of device_link_add() for the given consumer/supplier pair has not passed DL_FLAG_STATELESS to it in flags, the driver core should track the status of the link and act on it as appropriate (ie. the link should be treated as "managed"). This means that DL_FLAG_STATELESS needs to be set for managed device links and it should be valid to call device_link_del() or device_link_remove() to drop references to them in certain sutiations. To allow that to happen, introduce a new (internal) device link flag called DL_FLAG_MANAGED and make device_link_add() set it automatically whenever DL_FLAG_STATELESS is not passed to it. Also make it take additional references to existing device links that were previously stateless (that is, with DL_FLAG_STATELESS set and DL_FLAG_MANAGED unset) and will need to be managed going forward and initialize their status (which has been DL_STATE_NONE so far). Accordingly, when a managed device link is dropped automatically by the driver core, make it clear DL_FLAG_MANAGED, reset the link's status back to DL_STATE_NONE and drop the reference to it associated with DL_FLAG_MANAGED instead of just deleting it right away (to allow it to stay around in case it still needs to be released explicitly by someone). With that, since setting DL_FLAG_STATELESS doesn't mean that the device link in question is not managed any more, replace all of the status-tracking checks against DL_FLAG_STATELESS with analogous checks against DL_FLAG_MANAGED and update the documentation to reflect these changes. While at it, make device_link_add() reject flags that it does not recognize, including DL_FLAG_MANAGED. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Reviewed-by: Saravana Kannan <saravanak@google.com> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Review-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/2305283.AStDPdUUnE@kreacher Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-07-16 15:21:06 +00:00
* Links without the DL_FLAG_MANAGED flag set are ignored.
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
*/
bool device_links_busy(struct device *dev)
{
struct device_link *link;
bool ret = false;
device_links_write_lock();
list_for_each_entry(link, &dev->links.consumers, s_node) {
driver core: Remove device link creation limitation If device_link_add() is called for a consumer/supplier pair with an existing device link between them and the existing link's type is not in agreement with the flags passed to that function by its caller, NULL will be returned. That is seriously inconvenient, because it forces the callers of device_link_add() to worry about what others may or may not do even if that is not relevant to them for any other reasons. It turns out, however, that this limitation can be made go away relatively easily. The underlying observation is that if DL_FLAG_STATELESS has been passed to device_link_add() in flags for the given consumer/supplier pair at least once, calling either device_link_del() or device_link_remove() to release the link returned by it should work, but there are no other requirements associated with that flag. In turn, if at least one of the callers of device_link_add() for the given consumer/supplier pair has not passed DL_FLAG_STATELESS to it in flags, the driver core should track the status of the link and act on it as appropriate (ie. the link should be treated as "managed"). This means that DL_FLAG_STATELESS needs to be set for managed device links and it should be valid to call device_link_del() or device_link_remove() to drop references to them in certain sutiations. To allow that to happen, introduce a new (internal) device link flag called DL_FLAG_MANAGED and make device_link_add() set it automatically whenever DL_FLAG_STATELESS is not passed to it. Also make it take additional references to existing device links that were previously stateless (that is, with DL_FLAG_STATELESS set and DL_FLAG_MANAGED unset) and will need to be managed going forward and initialize their status (which has been DL_STATE_NONE so far). Accordingly, when a managed device link is dropped automatically by the driver core, make it clear DL_FLAG_MANAGED, reset the link's status back to DL_STATE_NONE and drop the reference to it associated with DL_FLAG_MANAGED instead of just deleting it right away (to allow it to stay around in case it still needs to be released explicitly by someone). With that, since setting DL_FLAG_STATELESS doesn't mean that the device link in question is not managed any more, replace all of the status-tracking checks against DL_FLAG_STATELESS with analogous checks against DL_FLAG_MANAGED and update the documentation to reflect these changes. While at it, make device_link_add() reject flags that it does not recognize, including DL_FLAG_MANAGED. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Reviewed-by: Saravana Kannan <saravanak@google.com> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Review-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/2305283.AStDPdUUnE@kreacher Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-07-16 15:21:06 +00:00
if (!(link->flags & DL_FLAG_MANAGED))
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
continue;
if (link->status == DL_STATE_CONSUMER_PROBE
|| link->status == DL_STATE_ACTIVE) {
ret = true;
break;
}
WRITE_ONCE(link->status, DL_STATE_SUPPLIER_UNBIND);
}
dev->links.status = DL_DEV_UNBINDING;
device_links_write_unlock();
return ret;
}
/**
* device_links_unbind_consumers - Force unbind consumers of the given device.
* @dev: Device to unbind the consumers of.
*
* Walk the list of links to consumers for @dev and if any of them is in the
* "consumer probe" state, wait for all device probes in progress to complete
* and start over.
*
* If that's not the case, change the status of the link to "supplier unbind"
* and check if the link was in the "active" state. If so, force the consumer
* driver to unbind and start over (the consumer will not re-probe as we have
* changed the state of the link already).
*
driver core: Remove device link creation limitation If device_link_add() is called for a consumer/supplier pair with an existing device link between them and the existing link's type is not in agreement with the flags passed to that function by its caller, NULL will be returned. That is seriously inconvenient, because it forces the callers of device_link_add() to worry about what others may or may not do even if that is not relevant to them for any other reasons. It turns out, however, that this limitation can be made go away relatively easily. The underlying observation is that if DL_FLAG_STATELESS has been passed to device_link_add() in flags for the given consumer/supplier pair at least once, calling either device_link_del() or device_link_remove() to release the link returned by it should work, but there are no other requirements associated with that flag. In turn, if at least one of the callers of device_link_add() for the given consumer/supplier pair has not passed DL_FLAG_STATELESS to it in flags, the driver core should track the status of the link and act on it as appropriate (ie. the link should be treated as "managed"). This means that DL_FLAG_STATELESS needs to be set for managed device links and it should be valid to call device_link_del() or device_link_remove() to drop references to them in certain sutiations. To allow that to happen, introduce a new (internal) device link flag called DL_FLAG_MANAGED and make device_link_add() set it automatically whenever DL_FLAG_STATELESS is not passed to it. Also make it take additional references to existing device links that were previously stateless (that is, with DL_FLAG_STATELESS set and DL_FLAG_MANAGED unset) and will need to be managed going forward and initialize their status (which has been DL_STATE_NONE so far). Accordingly, when a managed device link is dropped automatically by the driver core, make it clear DL_FLAG_MANAGED, reset the link's status back to DL_STATE_NONE and drop the reference to it associated with DL_FLAG_MANAGED instead of just deleting it right away (to allow it to stay around in case it still needs to be released explicitly by someone). With that, since setting DL_FLAG_STATELESS doesn't mean that the device link in question is not managed any more, replace all of the status-tracking checks against DL_FLAG_STATELESS with analogous checks against DL_FLAG_MANAGED and update the documentation to reflect these changes. While at it, make device_link_add() reject flags that it does not recognize, including DL_FLAG_MANAGED. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Reviewed-by: Saravana Kannan <saravanak@google.com> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Review-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/2305283.AStDPdUUnE@kreacher Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-07-16 15:21:06 +00:00
* Links without the DL_FLAG_MANAGED flag set are ignored.
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
*/
void device_links_unbind_consumers(struct device *dev)
{
struct device_link *link;
start:
device_links_write_lock();
list_for_each_entry(link, &dev->links.consumers, s_node) {
enum device_link_state status;
if (!(link->flags & DL_FLAG_MANAGED) ||
link->flags & DL_FLAG_SYNC_STATE_ONLY)
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
continue;
status = link->status;
if (status == DL_STATE_CONSUMER_PROBE) {
device_links_write_unlock();
wait_for_device_probe();
goto start;
}
WRITE_ONCE(link->status, DL_STATE_SUPPLIER_UNBIND);
if (status == DL_STATE_ACTIVE) {
struct device *consumer = link->consumer;
get_device(consumer);
device_links_write_unlock();
device_release_driver_internal(consumer, NULL,
consumer->parent);
put_device(consumer);
goto start;
}
}
device_links_write_unlock();
}
/**
* device_links_purge - Delete existing links to other devices.
* @dev: Target device.
*/
static void device_links_purge(struct device *dev)
{
struct device_link *link, *ln;
if (dev->class == &devlink_class)
return;
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
/*
* Delete all of the remaining links from this device to any other
* devices (either consumers or suppliers).
*/
device_links_write_lock();
list_for_each_entry_safe_reverse(link, ln, &dev->links.suppliers, c_node) {
WARN_ON(link->status == DL_STATE_ACTIVE);
__device_link_del(&link->kref);
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
}
list_for_each_entry_safe_reverse(link, ln, &dev->links.consumers, s_node) {
WARN_ON(link->status != DL_STATE_DORMANT &&
link->status != DL_STATE_NONE);
__device_link_del(&link->kref);
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
}
device_links_write_unlock();
}
#define FW_DEVLINK_FLAGS_PERMISSIVE (DL_FLAG_INFERRED | \
DL_FLAG_SYNC_STATE_ONLY)
#define FW_DEVLINK_FLAGS_ON (DL_FLAG_INFERRED | \
DL_FLAG_AUTOPROBE_CONSUMER)
#define FW_DEVLINK_FLAGS_RPM (FW_DEVLINK_FLAGS_ON | \
DL_FLAG_PM_RUNTIME)
static u32 fw_devlink_flags = FW_DEVLINK_FLAGS_ON;
static int __init fw_devlink_setup(char *arg)
{
if (!arg)
return -EINVAL;
if (strcmp(arg, "off") == 0) {
fw_devlink_flags = 0;
} else if (strcmp(arg, "permissive") == 0) {
fw_devlink_flags = FW_DEVLINK_FLAGS_PERMISSIVE;
} else if (strcmp(arg, "on") == 0) {
fw_devlink_flags = FW_DEVLINK_FLAGS_ON;
} else if (strcmp(arg, "rpm") == 0) {
fw_devlink_flags = FW_DEVLINK_FLAGS_RPM;
}
return 0;
}
early_param("fw_devlink", fw_devlink_setup);
static bool fw_devlink_strict;
static int __init fw_devlink_strict_setup(char *arg)
{
return strtobool(arg, &fw_devlink_strict);
}
early_param("fw_devlink.strict", fw_devlink_strict_setup);
u32 fw_devlink_get_flags(void)
{
return fw_devlink_flags;
}
static bool fw_devlink_is_permissive(void)
{
return fw_devlink_flags == FW_DEVLINK_FLAGS_PERMISSIVE;
}
bool fw_devlink_is_strict(void)
{
return fw_devlink_strict && !fw_devlink_is_permissive();
}
static void fw_devlink_parse_fwnode(struct fwnode_handle *fwnode)
{
if (fwnode->flags & FWNODE_FLAG_LINKS_ADDED)
return;
fwnode_call_int_op(fwnode, add_links);
fwnode->flags |= FWNODE_FLAG_LINKS_ADDED;
}
static void fw_devlink_parse_fwtree(struct fwnode_handle *fwnode)
{
struct fwnode_handle *child = NULL;
fw_devlink_parse_fwnode(fwnode);
while ((child = fwnode_get_next_available_child_node(fwnode, child)))
fw_devlink_parse_fwtree(child);
}
driver core: Improve fw_devlink & deferred_probe_timeout interaction deferred_probe_timeout kernel commandline parameter allows probing of consumer devices if the supplier devices don't have any drivers. fw_devlink=on will indefintely block probe() calls on a device if all its suppliers haven't probed successfully. This completely skips calls to driver_deferred_probe_check_state() since that's only called when a .probe() function calls framework APIs. So fw_devlink=on breaks deferred_probe_timeout. deferred_probe_timeout in its current state also ignores a lot of information that's now available to the kernel. It assumes all suppliers that haven't probed when the timer expires (or when initcalls are done on a static kernel) will never probe and fails any calls to acquire resources from these unprobed suppliers. However, this assumption by deferred_probe_timeout isn't true under many conditions. For example: - If the consumer happens to be before the supplier in the deferred probe list. - If the supplier itself is waiting on its supplier to probe. This patch fixes both these issues by relaxing device links between devices only if the supplier doesn't have any driver that could match with (NOT bound to) the supplier device. This way, we only fail attempts to acquire resources from suppliers that truly don't have any driver vs suppliers that just happen to not have probed yet. Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20210402040342.2944858-3-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2021-04-02 04:03:41 +00:00
static void fw_devlink_relax_link(struct device_link *link)
{
if (!(link->flags & DL_FLAG_INFERRED))
return;
if (link->flags == (DL_FLAG_MANAGED | FW_DEVLINK_FLAGS_PERMISSIVE))
return;
pm_runtime_drop_link(link);
link->flags = DL_FLAG_MANAGED | FW_DEVLINK_FLAGS_PERMISSIVE;
dev_dbg(link->consumer, "Relaxing link with %s\n",
dev_name(link->supplier));
}
static int fw_devlink_no_driver(struct device *dev, void *data)
{
struct device_link *link = to_devlink(dev);
if (!link->supplier->can_match)
fw_devlink_relax_link(link);
return 0;
}
void fw_devlink_drivers_done(void)
{
fw_devlink_drv_reg_done = true;
device_links_write_lock();
class_for_each_device(&devlink_class, NULL, NULL,
fw_devlink_no_driver);
device_links_write_unlock();
}
driver core: Add wait_for_init_devices_probe helper function Some devices might need to be probed and bound successfully before the kernel boot sequence can finish and move on to init/userspace. For example, a network interface might need to be bound to be able to mount a NFS rootfs. With fw_devlink=on by default, some of these devices might be blocked from probing because they are waiting on a optional supplier that doesn't have a driver. While fw_devlink will eventually identify such devices and unblock the probing automatically, it might be too late by the time it unblocks the probing of devices. For example, the IP4 autoconfig might timeout before fw_devlink unblocks probing of the network interface. This function is available to temporarily try and probe all devices that have a driver even if some of their suppliers haven't been added or don't have drivers. The drivers can then decide which of the suppliers are optional vs mandatory and probe the device if possible. By the time this function returns, all such "best effort" probes are guaranteed to be completed. If a device successfully probes in this mode, we delete all fw_devlink discovered dependencies of that device where the supplier hasn't yet probed successfully because they have to be optional dependencies. This also means that some devices that aren't needed for init and could have waited for their optional supplier to probe (when the supplier's module is loaded later on) would end up probing prematurely with limited functionality. So call this function only when boot would fail without it. Tested-by: Geert Uytterhoeven <geert+renesas@glider.be> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20220601070707.3946847-5-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2022-06-01 07:07:00 +00:00
/**
* wait_for_init_devices_probe - Try to probe any device needed for init
*
* Some devices might need to be probed and bound successfully before the kernel
* boot sequence can finish and move on to init/userspace. For example, a
* network interface might need to be bound to be able to mount a NFS rootfs.
*
* With fw_devlink=on by default, some of these devices might be blocked from
* probing because they are waiting on a optional supplier that doesn't have a
* driver. While fw_devlink will eventually identify such devices and unblock
* the probing automatically, it might be too late by the time it unblocks the
* probing of devices. For example, the IP4 autoconfig might timeout before
* fw_devlink unblocks probing of the network interface.
*
* This function is available to temporarily try and probe all devices that have
* a driver even if some of their suppliers haven't been added or don't have
* drivers.
*
* The drivers can then decide which of the suppliers are optional vs mandatory
* and probe the device if possible. By the time this function returns, all such
* "best effort" probes are guaranteed to be completed. If a device successfully
* probes in this mode, we delete all fw_devlink discovered dependencies of that
* device where the supplier hasn't yet probed successfully because they have to
* be optional dependencies.
*
* Any devices that didn't successfully probe go back to being treated as if
* this function was never called.
*
* This also means that some devices that aren't needed for init and could have
* waited for their optional supplier to probe (when the supplier's module is
* loaded later on) would end up probing prematurely with limited functionality.
* So call this function only when boot would fail without it.
*/
void __init wait_for_init_devices_probe(void)
{
if (!fw_devlink_flags || fw_devlink_is_permissive())
return;
/*
* Wait for all ongoing probes to finish so that the "best effort" is
* only applied to devices that can't probe otherwise.
*/
wait_for_device_probe();
pr_info("Trying to probe devices needed for running init ...\n");
fw_devlink_best_effort = true;
driver_deferred_probe_trigger();
/*
* Wait for all "best effort" probes to finish before going back to
* normal enforcement.
*/
wait_for_device_probe();
fw_devlink_best_effort = false;
}
driver core: Improve fw_devlink & deferred_probe_timeout interaction deferred_probe_timeout kernel commandline parameter allows probing of consumer devices if the supplier devices don't have any drivers. fw_devlink=on will indefintely block probe() calls on a device if all its suppliers haven't probed successfully. This completely skips calls to driver_deferred_probe_check_state() since that's only called when a .probe() function calls framework APIs. So fw_devlink=on breaks deferred_probe_timeout. deferred_probe_timeout in its current state also ignores a lot of information that's now available to the kernel. It assumes all suppliers that haven't probed when the timer expires (or when initcalls are done on a static kernel) will never probe and fails any calls to acquire resources from these unprobed suppliers. However, this assumption by deferred_probe_timeout isn't true under many conditions. For example: - If the consumer happens to be before the supplier in the deferred probe list. - If the supplier itself is waiting on its supplier to probe. This patch fixes both these issues by relaxing device links between devices only if the supplier doesn't have any driver that could match with (NOT bound to) the supplier device. This way, we only fail attempts to acquire resources from suppliers that truly don't have any driver vs suppliers that just happen to not have probed yet. Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20210402040342.2944858-3-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2021-04-02 04:03:41 +00:00
static void fw_devlink_unblock_consumers(struct device *dev)
{
struct device_link *link;
if (!fw_devlink_flags || fw_devlink_is_permissive())
return;
device_links_write_lock();
list_for_each_entry(link, &dev->links.consumers, s_node)
fw_devlink_relax_link(link);
device_links_write_unlock();
}
/**
* fw_devlink_relax_cycle - Convert cyclic links to SYNC_STATE_ONLY links
* @con: Device to check dependencies for.
* @sup: Device to check against.
*
* Check if @sup depends on @con or any device dependent on it (its child or
* its consumer etc). When such a cyclic dependency is found, convert all
* device links created solely by fw_devlink into SYNC_STATE_ONLY device links.
* This is the equivalent of doing fw_devlink=permissive just between the
* devices in the cycle. We need to do this because, at this point, fw_devlink
* can't tell which of these dependencies is not a real dependency.
*
* Return 1 if a cycle is found. Otherwise, return 0.
*/
static int fw_devlink_relax_cycle(struct device *con, void *sup)
{
struct device_link *link;
int ret;
if (con == sup)
return 1;
ret = device_for_each_child(con, sup, fw_devlink_relax_cycle);
if (ret)
return ret;
list_for_each_entry(link, &con->links.consumers, s_node) {
if ((link->flags & ~DL_FLAG_INFERRED) ==
(DL_FLAG_SYNC_STATE_ONLY | DL_FLAG_MANAGED))
continue;
if (!fw_devlink_relax_cycle(link->consumer, sup))
continue;
ret = 1;
driver core: Improve fw_devlink & deferred_probe_timeout interaction deferred_probe_timeout kernel commandline parameter allows probing of consumer devices if the supplier devices don't have any drivers. fw_devlink=on will indefintely block probe() calls on a device if all its suppliers haven't probed successfully. This completely skips calls to driver_deferred_probe_check_state() since that's only called when a .probe() function calls framework APIs. So fw_devlink=on breaks deferred_probe_timeout. deferred_probe_timeout in its current state also ignores a lot of information that's now available to the kernel. It assumes all suppliers that haven't probed when the timer expires (or when initcalls are done on a static kernel) will never probe and fails any calls to acquire resources from these unprobed suppliers. However, this assumption by deferred_probe_timeout isn't true under many conditions. For example: - If the consumer happens to be before the supplier in the deferred probe list. - If the supplier itself is waiting on its supplier to probe. This patch fixes both these issues by relaxing device links between devices only if the supplier doesn't have any driver that could match with (NOT bound to) the supplier device. This way, we only fail attempts to acquire resources from suppliers that truly don't have any driver vs suppliers that just happen to not have probed yet. Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20210402040342.2944858-3-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2021-04-02 04:03:41 +00:00
fw_devlink_relax_link(link);
}
return ret;
}
driver core: Refactor fw_devlink feature The current implementation of fw_devlink is very inefficient because it tries to get away without creating fwnode links in the name of saving memory usage. Past attempts to optimize runtime at the cost of memory usage were blocked with request for data showing that the optimization made significant improvement for real world scenarios. We have those scenarios now. There have been several reports of boot time increase in the order of seconds in this thread [1]. Several OEMs and SoC manufacturers have also privately reported significant (350-400ms) increase in boot time due to all the parsing done by fw_devlink. So this patch uses all the setup done by the previous patches in this series to refactor fw_devlink to be more efficient. Most of the code has been moved out of firmware specific (DT mostly) code into driver core. This brings the following benefits: - Instead of parsing the device tree multiple times during bootup, fw_devlink parses each fwnode node/property only once and creates fwnode links. The rest of the fw_devlink code then just looks at these fwnode links to do rest of the work. - Makes it much easier to debug probe issue due to fw_devlink in the future. fw_devlink=on blocks the probing of devices if they depend on a device that hasn't been added yet. With this refactor, it'll be very easy to tell what that device is because we now have a reference to the fwnode of the device. - Much easier to add fw_devlink support to ACPI and other firmware types. A refactor to move the common bits from DT specific code to driver core was in my TODO list as a prerequisite to adding ACPI support to fw_devlink. This series gets that done. [1] - https://lore.kernel.org/linux-omap/ea02f57e-871d-cd16-4418-c1da4bbc4696@ti.com/ Tested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Tested-by: Grygorii Strashko <grygorii.strashko@ti.com> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20201121020232.908850-17-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-11-21 02:02:31 +00:00
/**
* fw_devlink_create_devlink - Create a device link from a consumer to fwnode
* @con: consumer device for the device link
* @sup_handle: fwnode handle of supplier
* @flags: devlink flags
driver core: Refactor fw_devlink feature The current implementation of fw_devlink is very inefficient because it tries to get away without creating fwnode links in the name of saving memory usage. Past attempts to optimize runtime at the cost of memory usage were blocked with request for data showing that the optimization made significant improvement for real world scenarios. We have those scenarios now. There have been several reports of boot time increase in the order of seconds in this thread [1]. Several OEMs and SoC manufacturers have also privately reported significant (350-400ms) increase in boot time due to all the parsing done by fw_devlink. So this patch uses all the setup done by the previous patches in this series to refactor fw_devlink to be more efficient. Most of the code has been moved out of firmware specific (DT mostly) code into driver core. This brings the following benefits: - Instead of parsing the device tree multiple times during bootup, fw_devlink parses each fwnode node/property only once and creates fwnode links. The rest of the fw_devlink code then just looks at these fwnode links to do rest of the work. - Makes it much easier to debug probe issue due to fw_devlink in the future. fw_devlink=on blocks the probing of devices if they depend on a device that hasn't been added yet. With this refactor, it'll be very easy to tell what that device is because we now have a reference to the fwnode of the device. - Much easier to add fw_devlink support to ACPI and other firmware types. A refactor to move the common bits from DT specific code to driver core was in my TODO list as a prerequisite to adding ACPI support to fw_devlink. This series gets that done. [1] - https://lore.kernel.org/linux-omap/ea02f57e-871d-cd16-4418-c1da4bbc4696@ti.com/ Tested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Tested-by: Grygorii Strashko <grygorii.strashko@ti.com> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20201121020232.908850-17-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-11-21 02:02:31 +00:00
*
* This function will try to create a device link between the consumer device
* @con and the supplier device represented by @sup_handle.
*
* The supplier has to be provided as a fwnode because incorrect cycles in
* fwnode links can sometimes cause the supplier device to never be created.
* This function detects such cases and returns an error if it cannot create a
* device link from the consumer to a missing supplier.
*
* Returns,
* 0 on successfully creating a device link
* -EINVAL if the device link cannot be created as expected
* -EAGAIN if the device link cannot be created right now, but it may be
* possible to do that in the future
*/
static int fw_devlink_create_devlink(struct device *con,
struct fwnode_handle *sup_handle, u32 flags)
{
driver core: Refactor fw_devlink feature The current implementation of fw_devlink is very inefficient because it tries to get away without creating fwnode links in the name of saving memory usage. Past attempts to optimize runtime at the cost of memory usage were blocked with request for data showing that the optimization made significant improvement for real world scenarios. We have those scenarios now. There have been several reports of boot time increase in the order of seconds in this thread [1]. Several OEMs and SoC manufacturers have also privately reported significant (350-400ms) increase in boot time due to all the parsing done by fw_devlink. So this patch uses all the setup done by the previous patches in this series to refactor fw_devlink to be more efficient. Most of the code has been moved out of firmware specific (DT mostly) code into driver core. This brings the following benefits: - Instead of parsing the device tree multiple times during bootup, fw_devlink parses each fwnode node/property only once and creates fwnode links. The rest of the fw_devlink code then just looks at these fwnode links to do rest of the work. - Makes it much easier to debug probe issue due to fw_devlink in the future. fw_devlink=on blocks the probing of devices if they depend on a device that hasn't been added yet. With this refactor, it'll be very easy to tell what that device is because we now have a reference to the fwnode of the device. - Much easier to add fw_devlink support to ACPI and other firmware types. A refactor to move the common bits from DT specific code to driver core was in my TODO list as a prerequisite to adding ACPI support to fw_devlink. This series gets that done. [1] - https://lore.kernel.org/linux-omap/ea02f57e-871d-cd16-4418-c1da4bbc4696@ti.com/ Tested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Tested-by: Grygorii Strashko <grygorii.strashko@ti.com> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20201121020232.908850-17-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-11-21 02:02:31 +00:00
struct device *sup_dev;
int ret = 0;
/*
* In some cases, a device P might also be a supplier to its child node
* C. However, this would defer the probe of C until the probe of P
* completes successfully. This is perfectly fine in the device driver
* model. device_add() doesn't guarantee probe completion of the device
* by the time it returns.
*
* However, there are a few drivers that assume C will finish probing
* as soon as it's added and before P finishes probing. So, we provide
* a flag to let fw_devlink know not to delay the probe of C until the
* probe of P completes successfully.
*
* When such a flag is set, we can't create device links where P is the
* supplier of C as that would delay the probe of C.
*/
if (sup_handle->flags & FWNODE_FLAG_NEEDS_CHILD_BOUND_ON_ADD &&
fwnode_is_ancestor_of(sup_handle, con->fwnode))
return -EINVAL;
driver core: Refactor fw_devlink feature The current implementation of fw_devlink is very inefficient because it tries to get away without creating fwnode links in the name of saving memory usage. Past attempts to optimize runtime at the cost of memory usage were blocked with request for data showing that the optimization made significant improvement for real world scenarios. We have those scenarios now. There have been several reports of boot time increase in the order of seconds in this thread [1]. Several OEMs and SoC manufacturers have also privately reported significant (350-400ms) increase in boot time due to all the parsing done by fw_devlink. So this patch uses all the setup done by the previous patches in this series to refactor fw_devlink to be more efficient. Most of the code has been moved out of firmware specific (DT mostly) code into driver core. This brings the following benefits: - Instead of parsing the device tree multiple times during bootup, fw_devlink parses each fwnode node/property only once and creates fwnode links. The rest of the fw_devlink code then just looks at these fwnode links to do rest of the work. - Makes it much easier to debug probe issue due to fw_devlink in the future. fw_devlink=on blocks the probing of devices if they depend on a device that hasn't been added yet. With this refactor, it'll be very easy to tell what that device is because we now have a reference to the fwnode of the device. - Much easier to add fw_devlink support to ACPI and other firmware types. A refactor to move the common bits from DT specific code to driver core was in my TODO list as a prerequisite to adding ACPI support to fw_devlink. This series gets that done. [1] - https://lore.kernel.org/linux-omap/ea02f57e-871d-cd16-4418-c1da4bbc4696@ti.com/ Tested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Tested-by: Grygorii Strashko <grygorii.strashko@ti.com> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20201121020232.908850-17-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-11-21 02:02:31 +00:00
sup_dev = get_dev_from_fwnode(sup_handle);
if (sup_dev) {
/*
* If it's one of those drivers that don't actually bind to
* their device using driver core, then don't wait on this
* supplier device indefinitely.
*/
if (sup_dev->links.status == DL_DEV_NO_DRIVER &&
sup_handle->flags & FWNODE_FLAG_INITIALIZED) {
ret = -EINVAL;
goto out;
}
driver core: Refactor fw_devlink feature The current implementation of fw_devlink is very inefficient because it tries to get away without creating fwnode links in the name of saving memory usage. Past attempts to optimize runtime at the cost of memory usage were blocked with request for data showing that the optimization made significant improvement for real world scenarios. We have those scenarios now. There have been several reports of boot time increase in the order of seconds in this thread [1]. Several OEMs and SoC manufacturers have also privately reported significant (350-400ms) increase in boot time due to all the parsing done by fw_devlink. So this patch uses all the setup done by the previous patches in this series to refactor fw_devlink to be more efficient. Most of the code has been moved out of firmware specific (DT mostly) code into driver core. This brings the following benefits: - Instead of parsing the device tree multiple times during bootup, fw_devlink parses each fwnode node/property only once and creates fwnode links. The rest of the fw_devlink code then just looks at these fwnode links to do rest of the work. - Makes it much easier to debug probe issue due to fw_devlink in the future. fw_devlink=on blocks the probing of devices if they depend on a device that hasn't been added yet. With this refactor, it'll be very easy to tell what that device is because we now have a reference to the fwnode of the device. - Much easier to add fw_devlink support to ACPI and other firmware types. A refactor to move the common bits from DT specific code to driver core was in my TODO list as a prerequisite to adding ACPI support to fw_devlink. This series gets that done. [1] - https://lore.kernel.org/linux-omap/ea02f57e-871d-cd16-4418-c1da4bbc4696@ti.com/ Tested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Tested-by: Grygorii Strashko <grygorii.strashko@ti.com> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20201121020232.908850-17-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-11-21 02:02:31 +00:00
/*
* If this fails, it is due to cycles in device links. Just
* give up on this link and treat it as invalid.
*/
if (!device_link_add(con, sup_dev, flags) &&
!(flags & DL_FLAG_SYNC_STATE_ONLY)) {
dev_info(con, "Fixing up cyclic dependency with %s\n",
dev_name(sup_dev));
device_links_write_lock();
fw_devlink_relax_cycle(con, sup_dev);
device_links_write_unlock();
device_link_add(con, sup_dev,
FW_DEVLINK_FLAGS_PERMISSIVE);
driver core: Refactor fw_devlink feature The current implementation of fw_devlink is very inefficient because it tries to get away without creating fwnode links in the name of saving memory usage. Past attempts to optimize runtime at the cost of memory usage were blocked with request for data showing that the optimization made significant improvement for real world scenarios. We have those scenarios now. There have been several reports of boot time increase in the order of seconds in this thread [1]. Several OEMs and SoC manufacturers have also privately reported significant (350-400ms) increase in boot time due to all the parsing done by fw_devlink. So this patch uses all the setup done by the previous patches in this series to refactor fw_devlink to be more efficient. Most of the code has been moved out of firmware specific (DT mostly) code into driver core. This brings the following benefits: - Instead of parsing the device tree multiple times during bootup, fw_devlink parses each fwnode node/property only once and creates fwnode links. The rest of the fw_devlink code then just looks at these fwnode links to do rest of the work. - Makes it much easier to debug probe issue due to fw_devlink in the future. fw_devlink=on blocks the probing of devices if they depend on a device that hasn't been added yet. With this refactor, it'll be very easy to tell what that device is because we now have a reference to the fwnode of the device. - Much easier to add fw_devlink support to ACPI and other firmware types. A refactor to move the common bits from DT specific code to driver core was in my TODO list as a prerequisite to adding ACPI support to fw_devlink. This series gets that done. [1] - https://lore.kernel.org/linux-omap/ea02f57e-871d-cd16-4418-c1da4bbc4696@ti.com/ Tested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Tested-by: Grygorii Strashko <grygorii.strashko@ti.com> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20201121020232.908850-17-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-11-21 02:02:31 +00:00
ret = -EINVAL;
}
driver core: Refactor fw_devlink feature The current implementation of fw_devlink is very inefficient because it tries to get away without creating fwnode links in the name of saving memory usage. Past attempts to optimize runtime at the cost of memory usage were blocked with request for data showing that the optimization made significant improvement for real world scenarios. We have those scenarios now. There have been several reports of boot time increase in the order of seconds in this thread [1]. Several OEMs and SoC manufacturers have also privately reported significant (350-400ms) increase in boot time due to all the parsing done by fw_devlink. So this patch uses all the setup done by the previous patches in this series to refactor fw_devlink to be more efficient. Most of the code has been moved out of firmware specific (DT mostly) code into driver core. This brings the following benefits: - Instead of parsing the device tree multiple times during bootup, fw_devlink parses each fwnode node/property only once and creates fwnode links. The rest of the fw_devlink code then just looks at these fwnode links to do rest of the work. - Makes it much easier to debug probe issue due to fw_devlink in the future. fw_devlink=on blocks the probing of devices if they depend on a device that hasn't been added yet. With this refactor, it'll be very easy to tell what that device is because we now have a reference to the fwnode of the device. - Much easier to add fw_devlink support to ACPI and other firmware types. A refactor to move the common bits from DT specific code to driver core was in my TODO list as a prerequisite to adding ACPI support to fw_devlink. This series gets that done. [1] - https://lore.kernel.org/linux-omap/ea02f57e-871d-cd16-4418-c1da4bbc4696@ti.com/ Tested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Tested-by: Grygorii Strashko <grygorii.strashko@ti.com> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20201121020232.908850-17-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-11-21 02:02:31 +00:00
goto out;
}
driver core: Refactor fw_devlink feature The current implementation of fw_devlink is very inefficient because it tries to get away without creating fwnode links in the name of saving memory usage. Past attempts to optimize runtime at the cost of memory usage were blocked with request for data showing that the optimization made significant improvement for real world scenarios. We have those scenarios now. There have been several reports of boot time increase in the order of seconds in this thread [1]. Several OEMs and SoC manufacturers have also privately reported significant (350-400ms) increase in boot time due to all the parsing done by fw_devlink. So this patch uses all the setup done by the previous patches in this series to refactor fw_devlink to be more efficient. Most of the code has been moved out of firmware specific (DT mostly) code into driver core. This brings the following benefits: - Instead of parsing the device tree multiple times during bootup, fw_devlink parses each fwnode node/property only once and creates fwnode links. The rest of the fw_devlink code then just looks at these fwnode links to do rest of the work. - Makes it much easier to debug probe issue due to fw_devlink in the future. fw_devlink=on blocks the probing of devices if they depend on a device that hasn't been added yet. With this refactor, it'll be very easy to tell what that device is because we now have a reference to the fwnode of the device. - Much easier to add fw_devlink support to ACPI and other firmware types. A refactor to move the common bits from DT specific code to driver core was in my TODO list as a prerequisite to adding ACPI support to fw_devlink. This series gets that done. [1] - https://lore.kernel.org/linux-omap/ea02f57e-871d-cd16-4418-c1da4bbc4696@ti.com/ Tested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Tested-by: Grygorii Strashko <grygorii.strashko@ti.com> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20201121020232.908850-17-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-11-21 02:02:31 +00:00
/* Supplier that's already initialized without a struct device. */
if (sup_handle->flags & FWNODE_FLAG_INITIALIZED)
return -EINVAL;
driver core: Refactor fw_devlink feature The current implementation of fw_devlink is very inefficient because it tries to get away without creating fwnode links in the name of saving memory usage. Past attempts to optimize runtime at the cost of memory usage were blocked with request for data showing that the optimization made significant improvement for real world scenarios. We have those scenarios now. There have been several reports of boot time increase in the order of seconds in this thread [1]. Several OEMs and SoC manufacturers have also privately reported significant (350-400ms) increase in boot time due to all the parsing done by fw_devlink. So this patch uses all the setup done by the previous patches in this series to refactor fw_devlink to be more efficient. Most of the code has been moved out of firmware specific (DT mostly) code into driver core. This brings the following benefits: - Instead of parsing the device tree multiple times during bootup, fw_devlink parses each fwnode node/property only once and creates fwnode links. The rest of the fw_devlink code then just looks at these fwnode links to do rest of the work. - Makes it much easier to debug probe issue due to fw_devlink in the future. fw_devlink=on blocks the probing of devices if they depend on a device that hasn't been added yet. With this refactor, it'll be very easy to tell what that device is because we now have a reference to the fwnode of the device. - Much easier to add fw_devlink support to ACPI and other firmware types. A refactor to move the common bits from DT specific code to driver core was in my TODO list as a prerequisite to adding ACPI support to fw_devlink. This series gets that done. [1] - https://lore.kernel.org/linux-omap/ea02f57e-871d-cd16-4418-c1da4bbc4696@ti.com/ Tested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Tested-by: Grygorii Strashko <grygorii.strashko@ti.com> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20201121020232.908850-17-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-11-21 02:02:31 +00:00
/*
* DL_FLAG_SYNC_STATE_ONLY doesn't block probing and supports
* cycles. So cycle detection isn't necessary and shouldn't be
* done.
*/
if (flags & DL_FLAG_SYNC_STATE_ONLY)
return -EAGAIN;
/*
* If we can't find the supplier device from its fwnode, it might be
* due to a cyclic dependency between fwnodes. Some of these cycles can
* be broken by applying logic. Check for these types of cycles and
* break them so that devices in the cycle probe properly.
*
driver core: fw_devlink: Improve handling of cyclic dependencies When we have a dependency of the form: Device-A -> Device-C Device-B Device-C -> Device-B Where, * Indentation denotes "child of" parent in previous line. * X -> Y denotes X is consumer of Y based on firmware (Eg: DT). We have cyclic dependency: device-A -> device-C -> device-B -> device-A fw_devlink current treats device-C -> device-B dependency as an invalid dependency and doesn't enforce it but leaves the rest of the dependencies as is. While the current behavior is necessary, it is not sufficient if the false dependency in this example is actually device-A -> device-C. When this is the case, device-C will correctly probe defer waiting for device-B to be added, but device-A will be incorrectly probe deferred by fw_devlink waiting on device-C to probe successfully. Due to this, none of the devices in the cycle will end up probing. To fix this, we need to go relax all the dependencies in the cycle like we already do in the other instances where fw_devlink detects cycles. A real world example of this was reported[1] and analyzed[2]. [1] - https://lore.kernel.org/lkml/0a2c4106-7f48-2bb5-048e-8c001a7c3fda@samsung.com/ [2] - https://lore.kernel.org/lkml/CAGETcx8peaew90SWiux=TyvuGgvTQOmO4BFALz7aj0Za5QdNFQ@mail.gmail.com/ Fixes: f9aa460672c9 ("driver core: Refactor fw_devlink feature") Cc: stable <stable@vger.kernel.org> Reported-by: Marek Szyprowski <m.szyprowski@samsung.com> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20210915170940.617415-2-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2021-09-15 17:09:37 +00:00
* If the supplier's parent is dependent on the consumer, then the
* consumer and supplier have a cyclic dependency. Since fw_devlink
* can't tell which of the inferred dependencies are incorrect, don't
* enforce probe ordering between any of the devices in this cyclic
* dependency. Do this by relaxing all the fw_devlink device links in
* this cycle and by treating the fwnode link between the consumer and
* the supplier as an invalid dependency.
driver core: Refactor fw_devlink feature The current implementation of fw_devlink is very inefficient because it tries to get away without creating fwnode links in the name of saving memory usage. Past attempts to optimize runtime at the cost of memory usage were blocked with request for data showing that the optimization made significant improvement for real world scenarios. We have those scenarios now. There have been several reports of boot time increase in the order of seconds in this thread [1]. Several OEMs and SoC manufacturers have also privately reported significant (350-400ms) increase in boot time due to all the parsing done by fw_devlink. So this patch uses all the setup done by the previous patches in this series to refactor fw_devlink to be more efficient. Most of the code has been moved out of firmware specific (DT mostly) code into driver core. This brings the following benefits: - Instead of parsing the device tree multiple times during bootup, fw_devlink parses each fwnode node/property only once and creates fwnode links. The rest of the fw_devlink code then just looks at these fwnode links to do rest of the work. - Makes it much easier to debug probe issue due to fw_devlink in the future. fw_devlink=on blocks the probing of devices if they depend on a device that hasn't been added yet. With this refactor, it'll be very easy to tell what that device is because we now have a reference to the fwnode of the device. - Much easier to add fw_devlink support to ACPI and other firmware types. A refactor to move the common bits from DT specific code to driver core was in my TODO list as a prerequisite to adding ACPI support to fw_devlink. This series gets that done. [1] - https://lore.kernel.org/linux-omap/ea02f57e-871d-cd16-4418-c1da4bbc4696@ti.com/ Tested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Tested-by: Grygorii Strashko <grygorii.strashko@ti.com> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20201121020232.908850-17-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-11-21 02:02:31 +00:00
*/
sup_dev = fwnode_get_next_parent_dev(sup_handle);
if (sup_dev && device_is_dependent(con, sup_dev)) {
driver core: fw_devlink: Improve handling of cyclic dependencies When we have a dependency of the form: Device-A -> Device-C Device-B Device-C -> Device-B Where, * Indentation denotes "child of" parent in previous line. * X -> Y denotes X is consumer of Y based on firmware (Eg: DT). We have cyclic dependency: device-A -> device-C -> device-B -> device-A fw_devlink current treats device-C -> device-B dependency as an invalid dependency and doesn't enforce it but leaves the rest of the dependencies as is. While the current behavior is necessary, it is not sufficient if the false dependency in this example is actually device-A -> device-C. When this is the case, device-C will correctly probe defer waiting for device-B to be added, but device-A will be incorrectly probe deferred by fw_devlink waiting on device-C to probe successfully. Due to this, none of the devices in the cycle will end up probing. To fix this, we need to go relax all the dependencies in the cycle like we already do in the other instances where fw_devlink detects cycles. A real world example of this was reported[1] and analyzed[2]. [1] - https://lore.kernel.org/lkml/0a2c4106-7f48-2bb5-048e-8c001a7c3fda@samsung.com/ [2] - https://lore.kernel.org/lkml/CAGETcx8peaew90SWiux=TyvuGgvTQOmO4BFALz7aj0Za5QdNFQ@mail.gmail.com/ Fixes: f9aa460672c9 ("driver core: Refactor fw_devlink feature") Cc: stable <stable@vger.kernel.org> Reported-by: Marek Szyprowski <m.szyprowski@samsung.com> Tested-by: Marek Szyprowski <m.szyprowski@samsung.com> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20210915170940.617415-2-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2021-09-15 17:09:37 +00:00
dev_info(con, "Fixing up cyclic dependency with %pfwP (%s)\n",
sup_handle, dev_name(sup_dev));
device_links_write_lock();
fw_devlink_relax_cycle(con, sup_dev);
device_links_write_unlock();
driver core: Refactor fw_devlink feature The current implementation of fw_devlink is very inefficient because it tries to get away without creating fwnode links in the name of saving memory usage. Past attempts to optimize runtime at the cost of memory usage were blocked with request for data showing that the optimization made significant improvement for real world scenarios. We have those scenarios now. There have been several reports of boot time increase in the order of seconds in this thread [1]. Several OEMs and SoC manufacturers have also privately reported significant (350-400ms) increase in boot time due to all the parsing done by fw_devlink. So this patch uses all the setup done by the previous patches in this series to refactor fw_devlink to be more efficient. Most of the code has been moved out of firmware specific (DT mostly) code into driver core. This brings the following benefits: - Instead of parsing the device tree multiple times during bootup, fw_devlink parses each fwnode node/property only once and creates fwnode links. The rest of the fw_devlink code then just looks at these fwnode links to do rest of the work. - Makes it much easier to debug probe issue due to fw_devlink in the future. fw_devlink=on blocks the probing of devices if they depend on a device that hasn't been added yet. With this refactor, it'll be very easy to tell what that device is because we now have a reference to the fwnode of the device. - Much easier to add fw_devlink support to ACPI and other firmware types. A refactor to move the common bits from DT specific code to driver core was in my TODO list as a prerequisite to adding ACPI support to fw_devlink. This series gets that done. [1] - https://lore.kernel.org/linux-omap/ea02f57e-871d-cd16-4418-c1da4bbc4696@ti.com/ Tested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Tested-by: Grygorii Strashko <grygorii.strashko@ti.com> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20201121020232.908850-17-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-11-21 02:02:31 +00:00
ret = -EINVAL;
} else {
/*
* Can't check for cycles or no cycles. So let's try
* again later.
*/
ret = -EAGAIN;
}
out:
put_device(sup_dev);
return ret;
}
/**
* __fw_devlink_link_to_consumers - Create device links to consumers of a device
* @dev: Device that needs to be linked to its consumers
driver core: Refactor fw_devlink feature The current implementation of fw_devlink is very inefficient because it tries to get away without creating fwnode links in the name of saving memory usage. Past attempts to optimize runtime at the cost of memory usage were blocked with request for data showing that the optimization made significant improvement for real world scenarios. We have those scenarios now. There have been several reports of boot time increase in the order of seconds in this thread [1]. Several OEMs and SoC manufacturers have also privately reported significant (350-400ms) increase in boot time due to all the parsing done by fw_devlink. So this patch uses all the setup done by the previous patches in this series to refactor fw_devlink to be more efficient. Most of the code has been moved out of firmware specific (DT mostly) code into driver core. This brings the following benefits: - Instead of parsing the device tree multiple times during bootup, fw_devlink parses each fwnode node/property only once and creates fwnode links. The rest of the fw_devlink code then just looks at these fwnode links to do rest of the work. - Makes it much easier to debug probe issue due to fw_devlink in the future. fw_devlink=on blocks the probing of devices if they depend on a device that hasn't been added yet. With this refactor, it'll be very easy to tell what that device is because we now have a reference to the fwnode of the device. - Much easier to add fw_devlink support to ACPI and other firmware types. A refactor to move the common bits from DT specific code to driver core was in my TODO list as a prerequisite to adding ACPI support to fw_devlink. This series gets that done. [1] - https://lore.kernel.org/linux-omap/ea02f57e-871d-cd16-4418-c1da4bbc4696@ti.com/ Tested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Tested-by: Grygorii Strashko <grygorii.strashko@ti.com> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20201121020232.908850-17-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-11-21 02:02:31 +00:00
*
* This function looks at all the consumer fwnodes of @dev and creates device
* links between the consumer device and @dev (supplier).
*
* If the consumer device has not been added yet, then this function creates a
* SYNC_STATE_ONLY link between @dev (supplier) and the closest ancestor device
* of the consumer fwnode. This is necessary to make sure @dev doesn't get a
* sync_state() callback before the real consumer device gets to be added and
* then probed.
*
* Once device links are created from the real consumer to @dev (supplier), the
* fwnode links are deleted.
*/
static void __fw_devlink_link_to_consumers(struct device *dev)
{
struct fwnode_handle *fwnode = dev->fwnode;
struct fwnode_link *link, *tmp;
list_for_each_entry_safe(link, tmp, &fwnode->consumers, s_hook) {
u32 dl_flags = fw_devlink_get_flags();
struct device *con_dev;
bool own_link = true;
int ret;
con_dev = get_dev_from_fwnode(link->consumer);
/*
* If consumer device is not available yet, make a "proxy"
* SYNC_STATE_ONLY link from the consumer's parent device to
* the supplier device. This is necessary to make sure the
* supplier doesn't get a sync_state() callback before the real
* consumer can create a device link to the supplier.
*
* This proxy link step is needed to handle the case where the
* consumer's parent device is added before the supplier.
*/
if (!con_dev) {
con_dev = fwnode_get_next_parent_dev(link->consumer);
/*
* However, if the consumer's parent device is also the
* parent of the supplier, don't create a
* consumer-supplier link from the parent to its child
* device. Such a dependency is impossible.
*/
if (con_dev &&
fwnode_is_ancestor_of(con_dev->fwnode, fwnode)) {
put_device(con_dev);
con_dev = NULL;
} else {
own_link = false;
dl_flags = FW_DEVLINK_FLAGS_PERMISSIVE;
driver core: Refactor fw_devlink feature The current implementation of fw_devlink is very inefficient because it tries to get away without creating fwnode links in the name of saving memory usage. Past attempts to optimize runtime at the cost of memory usage were blocked with request for data showing that the optimization made significant improvement for real world scenarios. We have those scenarios now. There have been several reports of boot time increase in the order of seconds in this thread [1]. Several OEMs and SoC manufacturers have also privately reported significant (350-400ms) increase in boot time due to all the parsing done by fw_devlink. So this patch uses all the setup done by the previous patches in this series to refactor fw_devlink to be more efficient. Most of the code has been moved out of firmware specific (DT mostly) code into driver core. This brings the following benefits: - Instead of parsing the device tree multiple times during bootup, fw_devlink parses each fwnode node/property only once and creates fwnode links. The rest of the fw_devlink code then just looks at these fwnode links to do rest of the work. - Makes it much easier to debug probe issue due to fw_devlink in the future. fw_devlink=on blocks the probing of devices if they depend on a device that hasn't been added yet. With this refactor, it'll be very easy to tell what that device is because we now have a reference to the fwnode of the device. - Much easier to add fw_devlink support to ACPI and other firmware types. A refactor to move the common bits from DT specific code to driver core was in my TODO list as a prerequisite to adding ACPI support to fw_devlink. This series gets that done. [1] - https://lore.kernel.org/linux-omap/ea02f57e-871d-cd16-4418-c1da4bbc4696@ti.com/ Tested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Tested-by: Grygorii Strashko <grygorii.strashko@ti.com> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20201121020232.908850-17-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-11-21 02:02:31 +00:00
}
}
if (!con_dev)
continue;
ret = fw_devlink_create_devlink(con_dev, fwnode, dl_flags);
put_device(con_dev);
if (!own_link || ret == -EAGAIN)
continue;
__fwnode_link_del(link);
driver core: Refactor fw_devlink feature The current implementation of fw_devlink is very inefficient because it tries to get away without creating fwnode links in the name of saving memory usage. Past attempts to optimize runtime at the cost of memory usage were blocked with request for data showing that the optimization made significant improvement for real world scenarios. We have those scenarios now. There have been several reports of boot time increase in the order of seconds in this thread [1]. Several OEMs and SoC manufacturers have also privately reported significant (350-400ms) increase in boot time due to all the parsing done by fw_devlink. So this patch uses all the setup done by the previous patches in this series to refactor fw_devlink to be more efficient. Most of the code has been moved out of firmware specific (DT mostly) code into driver core. This brings the following benefits: - Instead of parsing the device tree multiple times during bootup, fw_devlink parses each fwnode node/property only once and creates fwnode links. The rest of the fw_devlink code then just looks at these fwnode links to do rest of the work. - Makes it much easier to debug probe issue due to fw_devlink in the future. fw_devlink=on blocks the probing of devices if they depend on a device that hasn't been added yet. With this refactor, it'll be very easy to tell what that device is because we now have a reference to the fwnode of the device. - Much easier to add fw_devlink support to ACPI and other firmware types. A refactor to move the common bits from DT specific code to driver core was in my TODO list as a prerequisite to adding ACPI support to fw_devlink. This series gets that done. [1] - https://lore.kernel.org/linux-omap/ea02f57e-871d-cd16-4418-c1da4bbc4696@ti.com/ Tested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Tested-by: Grygorii Strashko <grygorii.strashko@ti.com> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20201121020232.908850-17-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-11-21 02:02:31 +00:00
}
}
/**
* __fw_devlink_link_to_suppliers - Create device links to suppliers of a device
* @dev: The consumer device that needs to be linked to its suppliers
* @fwnode: Root of the fwnode tree that is used to create device links
driver core: Refactor fw_devlink feature The current implementation of fw_devlink is very inefficient because it tries to get away without creating fwnode links in the name of saving memory usage. Past attempts to optimize runtime at the cost of memory usage were blocked with request for data showing that the optimization made significant improvement for real world scenarios. We have those scenarios now. There have been several reports of boot time increase in the order of seconds in this thread [1]. Several OEMs and SoC manufacturers have also privately reported significant (350-400ms) increase in boot time due to all the parsing done by fw_devlink. So this patch uses all the setup done by the previous patches in this series to refactor fw_devlink to be more efficient. Most of the code has been moved out of firmware specific (DT mostly) code into driver core. This brings the following benefits: - Instead of parsing the device tree multiple times during bootup, fw_devlink parses each fwnode node/property only once and creates fwnode links. The rest of the fw_devlink code then just looks at these fwnode links to do rest of the work. - Makes it much easier to debug probe issue due to fw_devlink in the future. fw_devlink=on blocks the probing of devices if they depend on a device that hasn't been added yet. With this refactor, it'll be very easy to tell what that device is because we now have a reference to the fwnode of the device. - Much easier to add fw_devlink support to ACPI and other firmware types. A refactor to move the common bits from DT specific code to driver core was in my TODO list as a prerequisite to adding ACPI support to fw_devlink. This series gets that done. [1] - https://lore.kernel.org/linux-omap/ea02f57e-871d-cd16-4418-c1da4bbc4696@ti.com/ Tested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Tested-by: Grygorii Strashko <grygorii.strashko@ti.com> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20201121020232.908850-17-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-11-21 02:02:31 +00:00
*
* This function looks at all the supplier fwnodes of fwnode tree rooted at
* @fwnode and creates device links between @dev (consumer) and all the
* supplier devices of the entire fwnode tree at @fwnode.
*
* The function creates normal (non-SYNC_STATE_ONLY) device links between @dev
* and the real suppliers of @dev. Once these device links are created, the
* fwnode links are deleted. When such device links are successfully created,
* this function is called recursively on those supplier devices. This is
* needed to detect and break some invalid cycles in fwnode links. See
* fw_devlink_create_devlink() for more details.
*
* In addition, it also looks at all the suppliers of the entire fwnode tree
* because some of the child devices of @dev that have not been added yet
* (because @dev hasn't probed) might already have their suppliers added to
* driver core. So, this function creates SYNC_STATE_ONLY device links between
* @dev (consumer) and these suppliers to make sure they don't execute their
* sync_state() callbacks before these child devices have a chance to create
* their device links. The fwnode links that correspond to the child devices
* aren't delete because they are needed later to create the device links
* between the real consumer and supplier devices.
*/
static void __fw_devlink_link_to_suppliers(struct device *dev,
struct fwnode_handle *fwnode)
{
bool own_link = (dev->fwnode == fwnode);
struct fwnode_link *link, *tmp;
struct fwnode_handle *child = NULL;
u32 dl_flags;
if (own_link)
dl_flags = fw_devlink_get_flags();
else
dl_flags = FW_DEVLINK_FLAGS_PERMISSIVE;
driver core: Refactor fw_devlink feature The current implementation of fw_devlink is very inefficient because it tries to get away without creating fwnode links in the name of saving memory usage. Past attempts to optimize runtime at the cost of memory usage were blocked with request for data showing that the optimization made significant improvement for real world scenarios. We have those scenarios now. There have been several reports of boot time increase in the order of seconds in this thread [1]. Several OEMs and SoC manufacturers have also privately reported significant (350-400ms) increase in boot time due to all the parsing done by fw_devlink. So this patch uses all the setup done by the previous patches in this series to refactor fw_devlink to be more efficient. Most of the code has been moved out of firmware specific (DT mostly) code into driver core. This brings the following benefits: - Instead of parsing the device tree multiple times during bootup, fw_devlink parses each fwnode node/property only once and creates fwnode links. The rest of the fw_devlink code then just looks at these fwnode links to do rest of the work. - Makes it much easier to debug probe issue due to fw_devlink in the future. fw_devlink=on blocks the probing of devices if they depend on a device that hasn't been added yet. With this refactor, it'll be very easy to tell what that device is because we now have a reference to the fwnode of the device. - Much easier to add fw_devlink support to ACPI and other firmware types. A refactor to move the common bits from DT specific code to driver core was in my TODO list as a prerequisite to adding ACPI support to fw_devlink. This series gets that done. [1] - https://lore.kernel.org/linux-omap/ea02f57e-871d-cd16-4418-c1da4bbc4696@ti.com/ Tested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Tested-by: Grygorii Strashko <grygorii.strashko@ti.com> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20201121020232.908850-17-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-11-21 02:02:31 +00:00
list_for_each_entry_safe(link, tmp, &fwnode->suppliers, c_hook) {
int ret;
struct device *sup_dev;
struct fwnode_handle *sup = link->supplier;
ret = fw_devlink_create_devlink(dev, sup, dl_flags);
if (!own_link || ret == -EAGAIN)
continue;
__fwnode_link_del(link);
driver core: Refactor fw_devlink feature The current implementation of fw_devlink is very inefficient because it tries to get away without creating fwnode links in the name of saving memory usage. Past attempts to optimize runtime at the cost of memory usage were blocked with request for data showing that the optimization made significant improvement for real world scenarios. We have those scenarios now. There have been several reports of boot time increase in the order of seconds in this thread [1]. Several OEMs and SoC manufacturers have also privately reported significant (350-400ms) increase in boot time due to all the parsing done by fw_devlink. So this patch uses all the setup done by the previous patches in this series to refactor fw_devlink to be more efficient. Most of the code has been moved out of firmware specific (DT mostly) code into driver core. This brings the following benefits: - Instead of parsing the device tree multiple times during bootup, fw_devlink parses each fwnode node/property only once and creates fwnode links. The rest of the fw_devlink code then just looks at these fwnode links to do rest of the work. - Makes it much easier to debug probe issue due to fw_devlink in the future. fw_devlink=on blocks the probing of devices if they depend on a device that hasn't been added yet. With this refactor, it'll be very easy to tell what that device is because we now have a reference to the fwnode of the device. - Much easier to add fw_devlink support to ACPI and other firmware types. A refactor to move the common bits from DT specific code to driver core was in my TODO list as a prerequisite to adding ACPI support to fw_devlink. This series gets that done. [1] - https://lore.kernel.org/linux-omap/ea02f57e-871d-cd16-4418-c1da4bbc4696@ti.com/ Tested-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com> Tested-by: Grygorii Strashko <grygorii.strashko@ti.com> Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20201121020232.908850-17-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-11-21 02:02:31 +00:00
/* If no device link was created, nothing more to do. */
if (ret)
continue;
/*
* If a device link was successfully created to a supplier, we
* now need to try and link the supplier to all its suppliers.
*
* This is needed to detect and delete false dependencies in
* fwnode links that haven't been converted to a device link
* yet. See comments in fw_devlink_create_devlink() for more
* details on the false dependency.
*
* Without deleting these false dependencies, some devices will
* never probe because they'll keep waiting for their false
* dependency fwnode links to be converted to device links.
*/
sup_dev = get_dev_from_fwnode(sup);
__fw_devlink_link_to_suppliers(sup_dev, sup_dev->fwnode);
put_device(sup_dev);
}
/*
* Make "proxy" SYNC_STATE_ONLY device links to represent the needs of
* all the descendants. This proxy link step is needed to handle the
* case where the supplier is added before the consumer's parent device
* (@dev).
*/
while ((child = fwnode_get_next_available_child_node(fwnode, child)))
__fw_devlink_link_to_suppliers(dev, child);
}
static void fw_devlink_link_device(struct device *dev)
{
struct fwnode_handle *fwnode = dev->fwnode;
if (!fw_devlink_flags)
return;
fw_devlink_parse_fwtree(fwnode);
mutex_lock(&fwnode_link_lock);
__fw_devlink_link_to_consumers(dev);
__fw_devlink_link_to_suppliers(dev, fwnode);
mutex_unlock(&fwnode_link_lock);
}
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
/* Device links support end. */
int (*platform_notify)(struct device *dev) = NULL;
int (*platform_notify_remove)(struct device *dev) = NULL;
static struct kobject *dev_kobj;
struct kobject *sysfs_dev_char_kobj;
struct kobject *sysfs_dev_block_kobj;
driver core / ACPI: Avoid device hot remove locking issues device_hotplug_lock is held around the acpi_bus_trim() call in acpi_scan_hot_remove() which generally removes devices (it removes ACPI device objects at least, but it may also remove "physical" device objects through .detach() callbacks of ACPI scan handlers). Thus, potentially, device sysfs attributes are removed under that lock and to remove those attributes it is necessary to hold the s_active references of their directory entries for writing. On the other hand, the execution of a .show() or .store() callback from a sysfs attribute is carried out with that attribute's s_active reference held for reading. Consequently, if any device sysfs attribute that may be removed from within acpi_scan_hot_remove() through acpi_bus_trim() has a .store() or .show() callback which acquires device_hotplug_lock, the execution of that callback may deadlock with the removal of the attribute. [Unfortunately, the "online" device attribute of CPUs and memory blocks is one of them.] To avoid such deadlocks, make all of the sysfs attribute callbacks that need to lock device hotplug, for example store_online(), use a special function, lock_device_hotplug_sysfs(), to lock device hotplug and return the result of that function immediately if it is not zero. This will cause the s_active reference of the directory entry in question to be released and the syscall to be restarted if device_hotplug_lock cannot be acquired. [show_online() actually doesn't need to lock device hotplug, but it is useful to serialize it with respect to device_offline() and device_online() for the same device (in case user space attempts to run them concurrently) which can be done with the help of device_lock().] Reported-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com> Reported-and-tested-by: Gu Zheng <guz.fnst@cn.fujitsu.com> Suggested-by: Tejun Heo <tj@kernel.org> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Acked-by: Toshi Kani <toshi.kani@hp.com>
2013-08-28 19:41:01 +00:00
static DEFINE_MUTEX(device_hotplug_lock);
void lock_device_hotplug(void)
{
mutex_lock(&device_hotplug_lock);
}
void unlock_device_hotplug(void)
{
mutex_unlock(&device_hotplug_lock);
}
int lock_device_hotplug_sysfs(void)
{
if (mutex_trylock(&device_hotplug_lock))
return 0;
/* Avoid busy looping (5 ms of sleep should do). */
msleep(5);
return restart_syscall();
}
#ifdef CONFIG_BLOCK
static inline int device_is_not_partition(struct device *dev)
{
return !(dev->type == &part_type);
}
#else
static inline int device_is_not_partition(struct device *dev)
{
return 1;
}
#endif
static void device_platform_notify(struct device *dev)
{
acpi_device_notify(dev);
software_node_notify(dev);
if (platform_notify)
platform_notify(dev);
}
static void device_platform_notify_remove(struct device *dev)
{
acpi_device_notify_remove(dev);
software_node_notify_remove(dev);
if (platform_notify_remove)
platform_notify_remove(dev);
}
/**
* dev_driver_string - Return a device's driver name, if at all possible
* @dev: struct device to get the name of
*
* Will return the device's driver's name if it is bound to a device. If
* the device is not bound to a driver, it will return the name of the bus
* it is attached to. If it is not attached to a bus either, an empty
* string will be returned.
*/
const char *dev_driver_string(const struct device *dev)
{
struct device_driver *drv;
/* dev->driver can change to NULL underneath us because of unbinding,
* so be careful about accessing it. dev->bus and dev->class should
* never change once they are set, so they don't need special care.
*/
locking/atomics: COCCINELLE/treewide: Convert trivial ACCESS_ONCE() patterns to READ_ONCE()/WRITE_ONCE() Please do not apply this to mainline directly, instead please re-run the coccinelle script shown below and apply its output. For several reasons, it is desirable to use {READ,WRITE}_ONCE() in preference to ACCESS_ONCE(), and new code is expected to use one of the former. So far, there's been no reason to change most existing uses of ACCESS_ONCE(), as these aren't harmful, and changing them results in churn. However, for some features, the read/write distinction is critical to correct operation. To distinguish these cases, separate read/write accessors must be used. This patch migrates (most) remaining ACCESS_ONCE() instances to {READ,WRITE}_ONCE(), using the following coccinelle script: ---- // Convert trivial ACCESS_ONCE() uses to equivalent READ_ONCE() and // WRITE_ONCE() // $ make coccicheck COCCI=/home/mark/once.cocci SPFLAGS="--include-headers" MODE=patch virtual patch @ depends on patch @ expression E1, E2; @@ - ACCESS_ONCE(E1) = E2 + WRITE_ONCE(E1, E2) @ depends on patch @ expression E; @@ - ACCESS_ONCE(E) + READ_ONCE(E) ---- Signed-off-by: Mark Rutland <mark.rutland@arm.com> Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: davem@davemloft.net Cc: linux-arch@vger.kernel.org Cc: mpe@ellerman.id.au Cc: shuah@kernel.org Cc: snitzer@redhat.com Cc: thor.thayer@linux.intel.com Cc: tj@kernel.org Cc: viro@zeniv.linux.org.uk Cc: will.deacon@arm.com Link: http://lkml.kernel.org/r/1508792849-3115-19-git-send-email-paulmck@linux.vnet.ibm.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
2017-10-23 21:07:29 +00:00
drv = READ_ONCE(dev->driver);
return drv ? drv->name : dev_bus_name(dev);
}
EXPORT_SYMBOL(dev_driver_string);
#define to_dev_attr(_attr) container_of(_attr, struct device_attribute, attr)
static ssize_t dev_attr_show(struct kobject *kobj, struct attribute *attr,
char *buf)
{
struct device_attribute *dev_attr = to_dev_attr(attr);
struct device *dev = kobj_to_dev(kobj);
ssize_t ret = -EIO;
if (dev_attr->show)
ret = dev_attr->show(dev, dev_attr, buf);
if (ret >= (ssize_t)PAGE_SIZE) {
printk("dev_attr_show: %pS returned bad count\n",
dev_attr->show);
}
return ret;
}
static ssize_t dev_attr_store(struct kobject *kobj, struct attribute *attr,
const char *buf, size_t count)
{
struct device_attribute *dev_attr = to_dev_attr(attr);
struct device *dev = kobj_to_dev(kobj);
ssize_t ret = -EIO;
if (dev_attr->store)
ret = dev_attr->store(dev, dev_attr, buf, count);
return ret;
}
static const struct sysfs_ops dev_sysfs_ops = {
.show = dev_attr_show,
.store = dev_attr_store,
};
#define to_ext_attr(x) container_of(x, struct dev_ext_attribute, attr)
ssize_t device_store_ulong(struct device *dev,
struct device_attribute *attr,
const char *buf, size_t size)
{
struct dev_ext_attribute *ea = to_ext_attr(attr);
int ret;
unsigned long new;
ret = kstrtoul(buf, 0, &new);
if (ret)
return ret;
*(unsigned long *)(ea->var) = new;
/* Always return full write size even if we didn't consume all */
return size;
}
EXPORT_SYMBOL_GPL(device_store_ulong);
ssize_t device_show_ulong(struct device *dev,
struct device_attribute *attr,
char *buf)
{
struct dev_ext_attribute *ea = to_ext_attr(attr);
drivers core: Use sysfs_emit and sysfs_emit_at for show(device *...) functions Convert the various sprintf fmaily calls in sysfs device show functions to sysfs_emit and sysfs_emit_at for PAGE_SIZE buffer safety. Done with: $ spatch -sp-file sysfs_emit_dev.cocci --in-place --max-width=80 . And cocci script: $ cat sysfs_emit_dev.cocci @@ identifier d_show; identifier dev, attr, buf; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... return - sprintf(buf, + sysfs_emit(buf, ...); ...> } @@ identifier d_show; identifier dev, attr, buf; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... return - snprintf(buf, PAGE_SIZE, + sysfs_emit(buf, ...); ...> } @@ identifier d_show; identifier dev, attr, buf; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... return - scnprintf(buf, PAGE_SIZE, + sysfs_emit(buf, ...); ...> } @@ identifier d_show; identifier dev, attr, buf; expression chr; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... return - strcpy(buf, chr); + sysfs_emit(buf, chr); ...> } @@ identifier d_show; identifier dev, attr, buf; identifier len; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... len = - sprintf(buf, + sysfs_emit(buf, ...); ...> return len; } @@ identifier d_show; identifier dev, attr, buf; identifier len; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... len = - snprintf(buf, PAGE_SIZE, + sysfs_emit(buf, ...); ...> return len; } @@ identifier d_show; identifier dev, attr, buf; identifier len; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... len = - scnprintf(buf, PAGE_SIZE, + sysfs_emit(buf, ...); ...> return len; } @@ identifier d_show; identifier dev, attr, buf; identifier len; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... - len += scnprintf(buf + len, PAGE_SIZE - len, + len += sysfs_emit_at(buf, len, ...); ...> return len; } @@ identifier d_show; identifier dev, attr, buf; expression chr; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { ... - strcpy(buf, chr); - return strlen(buf); + return sysfs_emit(buf, chr); } Signed-off-by: Joe Perches <joe@perches.com> Link: https://lore.kernel.org/r/3d033c33056d88bbe34d4ddb62afd05ee166ab9a.1600285923.git.joe@perches.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-09-16 20:40:39 +00:00
return sysfs_emit(buf, "%lx\n", *(unsigned long *)(ea->var));
}
EXPORT_SYMBOL_GPL(device_show_ulong);
ssize_t device_store_int(struct device *dev,
struct device_attribute *attr,
const char *buf, size_t size)
{
struct dev_ext_attribute *ea = to_ext_attr(attr);
int ret;
long new;
ret = kstrtol(buf, 0, &new);
if (ret)
return ret;
if (new > INT_MAX || new < INT_MIN)
return -EINVAL;
*(int *)(ea->var) = new;
/* Always return full write size even if we didn't consume all */
return size;
}
EXPORT_SYMBOL_GPL(device_store_int);
ssize_t device_show_int(struct device *dev,
struct device_attribute *attr,
char *buf)
{
struct dev_ext_attribute *ea = to_ext_attr(attr);
drivers core: Use sysfs_emit and sysfs_emit_at for show(device *...) functions Convert the various sprintf fmaily calls in sysfs device show functions to sysfs_emit and sysfs_emit_at for PAGE_SIZE buffer safety. Done with: $ spatch -sp-file sysfs_emit_dev.cocci --in-place --max-width=80 . And cocci script: $ cat sysfs_emit_dev.cocci @@ identifier d_show; identifier dev, attr, buf; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... return - sprintf(buf, + sysfs_emit(buf, ...); ...> } @@ identifier d_show; identifier dev, attr, buf; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... return - snprintf(buf, PAGE_SIZE, + sysfs_emit(buf, ...); ...> } @@ identifier d_show; identifier dev, attr, buf; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... return - scnprintf(buf, PAGE_SIZE, + sysfs_emit(buf, ...); ...> } @@ identifier d_show; identifier dev, attr, buf; expression chr; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... return - strcpy(buf, chr); + sysfs_emit(buf, chr); ...> } @@ identifier d_show; identifier dev, attr, buf; identifier len; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... len = - sprintf(buf, + sysfs_emit(buf, ...); ...> return len; } @@ identifier d_show; identifier dev, attr, buf; identifier len; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... len = - snprintf(buf, PAGE_SIZE, + sysfs_emit(buf, ...); ...> return len; } @@ identifier d_show; identifier dev, attr, buf; identifier len; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... len = - scnprintf(buf, PAGE_SIZE, + sysfs_emit(buf, ...); ...> return len; } @@ identifier d_show; identifier dev, attr, buf; identifier len; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... - len += scnprintf(buf + len, PAGE_SIZE - len, + len += sysfs_emit_at(buf, len, ...); ...> return len; } @@ identifier d_show; identifier dev, attr, buf; expression chr; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { ... - strcpy(buf, chr); - return strlen(buf); + return sysfs_emit(buf, chr); } Signed-off-by: Joe Perches <joe@perches.com> Link: https://lore.kernel.org/r/3d033c33056d88bbe34d4ddb62afd05ee166ab9a.1600285923.git.joe@perches.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-09-16 20:40:39 +00:00
return sysfs_emit(buf, "%d\n", *(int *)(ea->var));
}
EXPORT_SYMBOL_GPL(device_show_int);
ssize_t device_store_bool(struct device *dev, struct device_attribute *attr,
const char *buf, size_t size)
{
struct dev_ext_attribute *ea = to_ext_attr(attr);
if (strtobool(buf, ea->var) < 0)
return -EINVAL;
return size;
}
EXPORT_SYMBOL_GPL(device_store_bool);
ssize_t device_show_bool(struct device *dev, struct device_attribute *attr,
char *buf)
{
struct dev_ext_attribute *ea = to_ext_attr(attr);
drivers core: Use sysfs_emit and sysfs_emit_at for show(device *...) functions Convert the various sprintf fmaily calls in sysfs device show functions to sysfs_emit and sysfs_emit_at for PAGE_SIZE buffer safety. Done with: $ spatch -sp-file sysfs_emit_dev.cocci --in-place --max-width=80 . And cocci script: $ cat sysfs_emit_dev.cocci @@ identifier d_show; identifier dev, attr, buf; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... return - sprintf(buf, + sysfs_emit(buf, ...); ...> } @@ identifier d_show; identifier dev, attr, buf; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... return - snprintf(buf, PAGE_SIZE, + sysfs_emit(buf, ...); ...> } @@ identifier d_show; identifier dev, attr, buf; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... return - scnprintf(buf, PAGE_SIZE, + sysfs_emit(buf, ...); ...> } @@ identifier d_show; identifier dev, attr, buf; expression chr; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... return - strcpy(buf, chr); + sysfs_emit(buf, chr); ...> } @@ identifier d_show; identifier dev, attr, buf; identifier len; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... len = - sprintf(buf, + sysfs_emit(buf, ...); ...> return len; } @@ identifier d_show; identifier dev, attr, buf; identifier len; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... len = - snprintf(buf, PAGE_SIZE, + sysfs_emit(buf, ...); ...> return len; } @@ identifier d_show; identifier dev, attr, buf; identifier len; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... len = - scnprintf(buf, PAGE_SIZE, + sysfs_emit(buf, ...); ...> return len; } @@ identifier d_show; identifier dev, attr, buf; identifier len; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... - len += scnprintf(buf + len, PAGE_SIZE - len, + len += sysfs_emit_at(buf, len, ...); ...> return len; } @@ identifier d_show; identifier dev, attr, buf; expression chr; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { ... - strcpy(buf, chr); - return strlen(buf); + return sysfs_emit(buf, chr); } Signed-off-by: Joe Perches <joe@perches.com> Link: https://lore.kernel.org/r/3d033c33056d88bbe34d4ddb62afd05ee166ab9a.1600285923.git.joe@perches.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-09-16 20:40:39 +00:00
return sysfs_emit(buf, "%d\n", *(bool *)(ea->var));
}
EXPORT_SYMBOL_GPL(device_show_bool);
/**
* device_release - free device structure.
* @kobj: device's kobject.
*
* This is called once the reference count for the object
* reaches 0. We forward the call to the device's release
* method, which should handle actually freeing the structure.
*/
static void device_release(struct kobject *kobj)
{
struct device *dev = kobj_to_dev(kobj);
struct device_private *p = dev->p;
/*
* Some platform devices are driven without driver attached
* and managed resources may have been acquired. Make sure
* all resources are released.
*
* Drivers still can add resources into device after device
* is deleted but alive, so release devres here to avoid
* possible memory leak.
*/
devres_release_all(dev);
kfree(dev->dma_range_map);
if (dev->release)
dev->release(dev);
else if (dev->type && dev->type->release)
dev->type->release(dev);
else if (dev->class && dev->class->dev_release)
dev->class->dev_release(dev);
else
WARN(1, KERN_ERR "Device '%s' does not have a release() function, it is broken and must be fixed. See Documentation/core-api/kobject.rst.\n",
dev_name(dev));
kfree(p);
}
static const void *device_namespace(struct kobject *kobj)
{
struct device *dev = kobj_to_dev(kobj);
const void *ns = NULL;
if (dev->class && dev->class->ns_type)
ns = dev->class->namespace(dev);
return ns;
}
static void device_get_ownership(struct kobject *kobj, kuid_t *uid, kgid_t *gid)
{
struct device *dev = kobj_to_dev(kobj);
if (dev->class && dev->class->get_ownership)
dev->class->get_ownership(dev, uid, gid);
}
static struct kobj_type device_ktype = {
.release = device_release,
.sysfs_ops = &dev_sysfs_ops,
.namespace = device_namespace,
.get_ownership = device_get_ownership,
};
static int dev_uevent_filter(struct kobject *kobj)
{
const struct kobj_type *ktype = get_ktype(kobj);
if (ktype == &device_ktype) {
struct device *dev = kobj_to_dev(kobj);
if (dev->bus)
return 1;
if (dev->class)
return 1;
}
return 0;
}
static const char *dev_uevent_name(struct kobject *kobj)
{
struct device *dev = kobj_to_dev(kobj);
if (dev->bus)
return dev->bus->name;
if (dev->class)
return dev->class->name;
return NULL;
}
static int dev_uevent(struct kobject *kobj, struct kobj_uevent_env *env)
{
struct device *dev = kobj_to_dev(kobj);
int retval = 0;
/* add device node properties if present */
if (MAJOR(dev->devt)) {
const char *tmp;
const char *name;
umode_t mode = 0;
kuid_t uid = GLOBAL_ROOT_UID;
kgid_t gid = GLOBAL_ROOT_GID;
add_uevent_var(env, "MAJOR=%u", MAJOR(dev->devt));
add_uevent_var(env, "MINOR=%u", MINOR(dev->devt));
name = device_get_devnode(dev, &mode, &uid, &gid, &tmp);
if (name) {
add_uevent_var(env, "DEVNAME=%s", name);
if (mode)
add_uevent_var(env, "DEVMODE=%#o", mode & 0777);
if (!uid_eq(uid, GLOBAL_ROOT_UID))
add_uevent_var(env, "DEVUID=%u", from_kuid(&init_user_ns, uid));
if (!gid_eq(gid, GLOBAL_ROOT_GID))
add_uevent_var(env, "DEVGID=%u", from_kgid(&init_user_ns, gid));
kfree(tmp);
}
}
if (dev->type && dev->type->name)
add_uevent_var(env, "DEVTYPE=%s", dev->type->name);
if (dev->driver)
add_uevent_var(env, "DRIVER=%s", dev->driver->name);
/* Add common DT information about the device */
of_device_uevent(dev, env);
/* have the bus specific function add its stuff */
if (dev->bus && dev->bus->uevent) {
retval = dev->bus->uevent(dev, env);
if (retval)
pr_debug("device: '%s': %s: bus uevent() returned %d\n",
dev_name(dev), __func__, retval);
}
/* have the class specific function add its stuff */
if (dev->class && dev->class->dev_uevent) {
retval = dev->class->dev_uevent(dev, env);
if (retval)
pr_debug("device: '%s': %s: class uevent() "
"returned %d\n", dev_name(dev),
__func__, retval);
}
/* have the device type specific function add its stuff */
if (dev->type && dev->type->uevent) {
retval = dev->type->uevent(dev, env);
if (retval)
pr_debug("device: '%s': %s: dev_type uevent() "
"returned %d\n", dev_name(dev),
__func__, retval);
}
return retval;
}
static const struct kset_uevent_ops device_uevent_ops = {
.filter = dev_uevent_filter,
.name = dev_uevent_name,
.uevent = dev_uevent,
};
static ssize_t uevent_show(struct device *dev, struct device_attribute *attr,
char *buf)
{
struct kobject *top_kobj;
struct kset *kset;
struct kobj_uevent_env *env = NULL;
int i;
int len = 0;
int retval;
/* search the kset, the device belongs to */
top_kobj = &dev->kobj;
while (!top_kobj->kset && top_kobj->parent)
top_kobj = top_kobj->parent;
if (!top_kobj->kset)
goto out;
kset = top_kobj->kset;
if (!kset->uevent_ops || !kset->uevent_ops->uevent)
goto out;
/* respect filter */
if (kset->uevent_ops && kset->uevent_ops->filter)
if (!kset->uevent_ops->filter(&dev->kobj))
goto out;
env = kzalloc(sizeof(struct kobj_uevent_env), GFP_KERNEL);
if (!env)
return -ENOMEM;
/* let the kset specific function add its keys */
retval = kset->uevent_ops->uevent(&dev->kobj, env);
if (retval)
goto out;
/* copy keys to file */
for (i = 0; i < env->envp_idx; i++)
len += sysfs_emit_at(buf, len, "%s\n", env->envp[i]);
out:
kfree(env);
return len;
}
static ssize_t uevent_store(struct device *dev, struct device_attribute *attr,
const char *buf, size_t count)
{
int rc;
rc = kobject_synth_uevent(&dev->kobj, buf, count);
if (rc) {
dev_err(dev, "uevent: failed to send synthetic uevent: %d\n", rc);
return rc;
}
return count;
}
static DEVICE_ATTR_RW(uevent);
static ssize_t online_show(struct device *dev, struct device_attribute *attr,
Driver core: Add offline/online device operations In some cases, graceful hot-removal of devices is not possible, although in principle the devices in question support hotplug. For example, that may happen for the last CPU in the system or for memory modules holding kernel memory. In those cases it is nice to be able to check if the given device can be gracefully hot-removed before triggering a removal procedure that cannot be aborted or reversed. Unfortunately, however, the kernel currently doesn't provide any support for that. To address that deficiency, introduce support for offline and online operations that can be performed on devices, respectively, before a hot-removal and in case when it is necessary (or convenient) to put a device back online after a successful offline (that has not been followed by removal). The idea is that the offline will fail whenever the given device cannot be gracefully removed from the system and it will not be allowed to use the device after a successful offline (until a subsequent online) in analogy with the existing CPU offline/online mechanism. For now, the offline and online operations are introduced at the bus type level, as that should be sufficient for the most urgent use cases (CPUs and memory modules). In the future, however, the approach may be extended to cover some more complicated device offline/online scenarios involving device drivers etc. The lock_device_hotplug() and unlock_device_hotplug() functions are introduced because subsequent patches need to put larger pieces of code under device_hotplug_lock to prevent race conditions between device offline and removal from happening. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Reviewed-by: Toshi Kani <toshi.kani@hp.com>
2013-05-02 20:15:29 +00:00
char *buf)
{
bool val;
driver core / ACPI: Avoid device hot remove locking issues device_hotplug_lock is held around the acpi_bus_trim() call in acpi_scan_hot_remove() which generally removes devices (it removes ACPI device objects at least, but it may also remove "physical" device objects through .detach() callbacks of ACPI scan handlers). Thus, potentially, device sysfs attributes are removed under that lock and to remove those attributes it is necessary to hold the s_active references of their directory entries for writing. On the other hand, the execution of a .show() or .store() callback from a sysfs attribute is carried out with that attribute's s_active reference held for reading. Consequently, if any device sysfs attribute that may be removed from within acpi_scan_hot_remove() through acpi_bus_trim() has a .store() or .show() callback which acquires device_hotplug_lock, the execution of that callback may deadlock with the removal of the attribute. [Unfortunately, the "online" device attribute of CPUs and memory blocks is one of them.] To avoid such deadlocks, make all of the sysfs attribute callbacks that need to lock device hotplug, for example store_online(), use a special function, lock_device_hotplug_sysfs(), to lock device hotplug and return the result of that function immediately if it is not zero. This will cause the s_active reference of the directory entry in question to be released and the syscall to be restarted if device_hotplug_lock cannot be acquired. [show_online() actually doesn't need to lock device hotplug, but it is useful to serialize it with respect to device_offline() and device_online() for the same device (in case user space attempts to run them concurrently) which can be done with the help of device_lock().] Reported-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com> Reported-and-tested-by: Gu Zheng <guz.fnst@cn.fujitsu.com> Suggested-by: Tejun Heo <tj@kernel.org> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Acked-by: Toshi Kani <toshi.kani@hp.com>
2013-08-28 19:41:01 +00:00
device_lock(dev);
Driver core: Add offline/online device operations In some cases, graceful hot-removal of devices is not possible, although in principle the devices in question support hotplug. For example, that may happen for the last CPU in the system or for memory modules holding kernel memory. In those cases it is nice to be able to check if the given device can be gracefully hot-removed before triggering a removal procedure that cannot be aborted or reversed. Unfortunately, however, the kernel currently doesn't provide any support for that. To address that deficiency, introduce support for offline and online operations that can be performed on devices, respectively, before a hot-removal and in case when it is necessary (or convenient) to put a device back online after a successful offline (that has not been followed by removal). The idea is that the offline will fail whenever the given device cannot be gracefully removed from the system and it will not be allowed to use the device after a successful offline (until a subsequent online) in analogy with the existing CPU offline/online mechanism. For now, the offline and online operations are introduced at the bus type level, as that should be sufficient for the most urgent use cases (CPUs and memory modules). In the future, however, the approach may be extended to cover some more complicated device offline/online scenarios involving device drivers etc. The lock_device_hotplug() and unlock_device_hotplug() functions are introduced because subsequent patches need to put larger pieces of code under device_hotplug_lock to prevent race conditions between device offline and removal from happening. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Reviewed-by: Toshi Kani <toshi.kani@hp.com>
2013-05-02 20:15:29 +00:00
val = !dev->offline;
driver core / ACPI: Avoid device hot remove locking issues device_hotplug_lock is held around the acpi_bus_trim() call in acpi_scan_hot_remove() which generally removes devices (it removes ACPI device objects at least, but it may also remove "physical" device objects through .detach() callbacks of ACPI scan handlers). Thus, potentially, device sysfs attributes are removed under that lock and to remove those attributes it is necessary to hold the s_active references of their directory entries for writing. On the other hand, the execution of a .show() or .store() callback from a sysfs attribute is carried out with that attribute's s_active reference held for reading. Consequently, if any device sysfs attribute that may be removed from within acpi_scan_hot_remove() through acpi_bus_trim() has a .store() or .show() callback which acquires device_hotplug_lock, the execution of that callback may deadlock with the removal of the attribute. [Unfortunately, the "online" device attribute of CPUs and memory blocks is one of them.] To avoid such deadlocks, make all of the sysfs attribute callbacks that need to lock device hotplug, for example store_online(), use a special function, lock_device_hotplug_sysfs(), to lock device hotplug and return the result of that function immediately if it is not zero. This will cause the s_active reference of the directory entry in question to be released and the syscall to be restarted if device_hotplug_lock cannot be acquired. [show_online() actually doesn't need to lock device hotplug, but it is useful to serialize it with respect to device_offline() and device_online() for the same device (in case user space attempts to run them concurrently) which can be done with the help of device_lock().] Reported-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com> Reported-and-tested-by: Gu Zheng <guz.fnst@cn.fujitsu.com> Suggested-by: Tejun Heo <tj@kernel.org> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Acked-by: Toshi Kani <toshi.kani@hp.com>
2013-08-28 19:41:01 +00:00
device_unlock(dev);
drivers core: Use sysfs_emit and sysfs_emit_at for show(device *...) functions Convert the various sprintf fmaily calls in sysfs device show functions to sysfs_emit and sysfs_emit_at for PAGE_SIZE buffer safety. Done with: $ spatch -sp-file sysfs_emit_dev.cocci --in-place --max-width=80 . And cocci script: $ cat sysfs_emit_dev.cocci @@ identifier d_show; identifier dev, attr, buf; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... return - sprintf(buf, + sysfs_emit(buf, ...); ...> } @@ identifier d_show; identifier dev, attr, buf; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... return - snprintf(buf, PAGE_SIZE, + sysfs_emit(buf, ...); ...> } @@ identifier d_show; identifier dev, attr, buf; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... return - scnprintf(buf, PAGE_SIZE, + sysfs_emit(buf, ...); ...> } @@ identifier d_show; identifier dev, attr, buf; expression chr; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... return - strcpy(buf, chr); + sysfs_emit(buf, chr); ...> } @@ identifier d_show; identifier dev, attr, buf; identifier len; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... len = - sprintf(buf, + sysfs_emit(buf, ...); ...> return len; } @@ identifier d_show; identifier dev, attr, buf; identifier len; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... len = - snprintf(buf, PAGE_SIZE, + sysfs_emit(buf, ...); ...> return len; } @@ identifier d_show; identifier dev, attr, buf; identifier len; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... len = - scnprintf(buf, PAGE_SIZE, + sysfs_emit(buf, ...); ...> return len; } @@ identifier d_show; identifier dev, attr, buf; identifier len; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { <... - len += scnprintf(buf + len, PAGE_SIZE - len, + len += sysfs_emit_at(buf, len, ...); ...> return len; } @@ identifier d_show; identifier dev, attr, buf; expression chr; @@ ssize_t d_show(struct device *dev, struct device_attribute *attr, char *buf) { ... - strcpy(buf, chr); - return strlen(buf); + return sysfs_emit(buf, chr); } Signed-off-by: Joe Perches <joe@perches.com> Link: https://lore.kernel.org/r/3d033c33056d88bbe34d4ddb62afd05ee166ab9a.1600285923.git.joe@perches.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2020-09-16 20:40:39 +00:00
return sysfs_emit(buf, "%u\n", val);
Driver core: Add offline/online device operations In some cases, graceful hot-removal of devices is not possible, although in principle the devices in question support hotplug. For example, that may happen for the last CPU in the system or for memory modules holding kernel memory. In those cases it is nice to be able to check if the given device can be gracefully hot-removed before triggering a removal procedure that cannot be aborted or reversed. Unfortunately, however, the kernel currently doesn't provide any support for that. To address that deficiency, introduce support for offline and online operations that can be performed on devices, respectively, before a hot-removal and in case when it is necessary (or convenient) to put a device back online after a successful offline (that has not been followed by removal). The idea is that the offline will fail whenever the given device cannot be gracefully removed from the system and it will not be allowed to use the device after a successful offline (until a subsequent online) in analogy with the existing CPU offline/online mechanism. For now, the offline and online operations are introduced at the bus type level, as that should be sufficient for the most urgent use cases (CPUs and memory modules). In the future, however, the approach may be extended to cover some more complicated device offline/online scenarios involving device drivers etc. The lock_device_hotplug() and unlock_device_hotplug() functions are introduced because subsequent patches need to put larger pieces of code under device_hotplug_lock to prevent race conditions between device offline and removal from happening. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Reviewed-by: Toshi Kani <toshi.kani@hp.com>
2013-05-02 20:15:29 +00:00
}
static ssize_t online_store(struct device *dev, struct device_attribute *attr,
Driver core: Add offline/online device operations In some cases, graceful hot-removal of devices is not possible, although in principle the devices in question support hotplug. For example, that may happen for the last CPU in the system or for memory modules holding kernel memory. In those cases it is nice to be able to check if the given device can be gracefully hot-removed before triggering a removal procedure that cannot be aborted or reversed. Unfortunately, however, the kernel currently doesn't provide any support for that. To address that deficiency, introduce support for offline and online operations that can be performed on devices, respectively, before a hot-removal and in case when it is necessary (or convenient) to put a device back online after a successful offline (that has not been followed by removal). The idea is that the offline will fail whenever the given device cannot be gracefully removed from the system and it will not be allowed to use the device after a successful offline (until a subsequent online) in analogy with the existing CPU offline/online mechanism. For now, the offline and online operations are introduced at the bus type level, as that should be sufficient for the most urgent use cases (CPUs and memory modules). In the future, however, the approach may be extended to cover some more complicated device offline/online scenarios involving device drivers etc. The lock_device_hotplug() and unlock_device_hotplug() functions are introduced because subsequent patches need to put larger pieces of code under device_hotplug_lock to prevent race conditions between device offline and removal from happening. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Reviewed-by: Toshi Kani <toshi.kani@hp.com>
2013-05-02 20:15:29 +00:00
const char *buf, size_t count)
{
bool val;
int ret;
ret = strtobool(buf, &val);
if (ret < 0)
return ret;
driver core / ACPI: Avoid device hot remove locking issues device_hotplug_lock is held around the acpi_bus_trim() call in acpi_scan_hot_remove() which generally removes devices (it removes ACPI device objects at least, but it may also remove "physical" device objects through .detach() callbacks of ACPI scan handlers). Thus, potentially, device sysfs attributes are removed under that lock and to remove those attributes it is necessary to hold the s_active references of their directory entries for writing. On the other hand, the execution of a .show() or .store() callback from a sysfs attribute is carried out with that attribute's s_active reference held for reading. Consequently, if any device sysfs attribute that may be removed from within acpi_scan_hot_remove() through acpi_bus_trim() has a .store() or .show() callback which acquires device_hotplug_lock, the execution of that callback may deadlock with the removal of the attribute. [Unfortunately, the "online" device attribute of CPUs and memory blocks is one of them.] To avoid such deadlocks, make all of the sysfs attribute callbacks that need to lock device hotplug, for example store_online(), use a special function, lock_device_hotplug_sysfs(), to lock device hotplug and return the result of that function immediately if it is not zero. This will cause the s_active reference of the directory entry in question to be released and the syscall to be restarted if device_hotplug_lock cannot be acquired. [show_online() actually doesn't need to lock device hotplug, but it is useful to serialize it with respect to device_offline() and device_online() for the same device (in case user space attempts to run them concurrently) which can be done with the help of device_lock().] Reported-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com> Reported-and-tested-by: Gu Zheng <guz.fnst@cn.fujitsu.com> Suggested-by: Tejun Heo <tj@kernel.org> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Acked-by: Toshi Kani <toshi.kani@hp.com>
2013-08-28 19:41:01 +00:00
ret = lock_device_hotplug_sysfs();
if (ret)
return ret;
Driver core: Add offline/online device operations In some cases, graceful hot-removal of devices is not possible, although in principle the devices in question support hotplug. For example, that may happen for the last CPU in the system or for memory modules holding kernel memory. In those cases it is nice to be able to check if the given device can be gracefully hot-removed before triggering a removal procedure that cannot be aborted or reversed. Unfortunately, however, the kernel currently doesn't provide any support for that. To address that deficiency, introduce support for offline and online operations that can be performed on devices, respectively, before a hot-removal and in case when it is necessary (or convenient) to put a device back online after a successful offline (that has not been followed by removal). The idea is that the offline will fail whenever the given device cannot be gracefully removed from the system and it will not be allowed to use the device after a successful offline (until a subsequent online) in analogy with the existing CPU offline/online mechanism. For now, the offline and online operations are introduced at the bus type level, as that should be sufficient for the most urgent use cases (CPUs and memory modules). In the future, however, the approach may be extended to cover some more complicated device offline/online scenarios involving device drivers etc. The lock_device_hotplug() and unlock_device_hotplug() functions are introduced because subsequent patches need to put larger pieces of code under device_hotplug_lock to prevent race conditions between device offline and removal from happening. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Reviewed-by: Toshi Kani <toshi.kani@hp.com>
2013-05-02 20:15:29 +00:00
ret = val ? device_online(dev) : device_offline(dev);
unlock_device_hotplug();
return ret < 0 ? ret : count;
}
static DEVICE_ATTR_RW(online);
Driver core: Add offline/online device operations In some cases, graceful hot-removal of devices is not possible, although in principle the devices in question support hotplug. For example, that may happen for the last CPU in the system or for memory modules holding kernel memory. In those cases it is nice to be able to check if the given device can be gracefully hot-removed before triggering a removal procedure that cannot be aborted or reversed. Unfortunately, however, the kernel currently doesn't provide any support for that. To address that deficiency, introduce support for offline and online operations that can be performed on devices, respectively, before a hot-removal and in case when it is necessary (or convenient) to put a device back online after a successful offline (that has not been followed by removal). The idea is that the offline will fail whenever the given device cannot be gracefully removed from the system and it will not be allowed to use the device after a successful offline (until a subsequent online) in analogy with the existing CPU offline/online mechanism. For now, the offline and online operations are introduced at the bus type level, as that should be sufficient for the most urgent use cases (CPUs and memory modules). In the future, however, the approach may be extended to cover some more complicated device offline/online scenarios involving device drivers etc. The lock_device_hotplug() and unlock_device_hotplug() functions are introduced because subsequent patches need to put larger pieces of code under device_hotplug_lock to prevent race conditions between device offline and removal from happening. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Reviewed-by: Toshi Kani <toshi.kani@hp.com>
2013-05-02 20:15:29 +00:00
driver core: Move the "removable" attribute from USB to core Move the "removable" attribute from USB to core in order to allow it to be supported by other subsystem / buses. Individual buses that want to support this attribute can populate the removable property of the device while enumerating it with the 3 possible values - - "unknown" - "fixed" - "removable" Leaving the field unchanged (i.e. "not supported") would mean that the attribute would not show up in sysfs for that device. The UAPI (location, symantics etc) for the attribute remains unchanged. Move the "removable" attribute from USB to the device core so it can be used by other subsystems / buses. By default, devices do not have a "removable" attribute in sysfs. If a subsystem or bus driver wants to support a "removable" attribute, it should call device_set_removable() before calling device_register() or device_add(), e.g.: device_set_removable(dev, DEVICE_REMOVABLE); device_register(dev); The possible values and the resulting sysfs attribute contents are: DEVICE_REMOVABLE_UNKNOWN -> "unknown" DEVICE_REMOVABLE -> "removable" DEVICE_FIXED -> "fixed" Convert the USB "removable" attribute to use this new device core functionality. There should be no user-visible change in the location or semantics of attribute for USB devices. Reviewed-by: Bjorn Helgaas <bhelgaas@google.com> Signed-off-by: Rajat Jain <rajatja@google.com> Link: https://lore.kernel.org/r/20210524171812.18095-1-rajatja@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2021-05-24 17:18:11 +00:00
static ssize_t removable_show(struct device *dev, struct device_attribute *attr,
char *buf)
{
const char *loc;
switch (dev->removable) {
case DEVICE_REMOVABLE:
loc = "removable";
break;
case DEVICE_FIXED:
loc = "fixed";
break;
default:
loc = "unknown";
}
return sysfs_emit(buf, "%s\n", loc);
}
static DEVICE_ATTR_RO(removable);
int device_add_groups(struct device *dev, const struct attribute_group **groups)
{
return sysfs_create_groups(&dev->kobj, groups);
}
EXPORT_SYMBOL_GPL(device_add_groups);
void device_remove_groups(struct device *dev,
const struct attribute_group **groups)
{
sysfs_remove_groups(&dev->kobj, groups);
}
EXPORT_SYMBOL_GPL(device_remove_groups);
union device_attr_group_devres {
const struct attribute_group *group;
const struct attribute_group **groups;
};
static int devm_attr_group_match(struct device *dev, void *res, void *data)
{
return ((union device_attr_group_devres *)res)->group == data;
}
static void devm_attr_group_remove(struct device *dev, void *res)
{
union device_attr_group_devres *devres = res;
const struct attribute_group *group = devres->group;
dev_dbg(dev, "%s: removing group %p\n", __func__, group);
sysfs_remove_group(&dev->kobj, group);
}
static void devm_attr_groups_remove(struct device *dev, void *res)
{
union device_attr_group_devres *devres = res;
const struct attribute_group **groups = devres->groups;
dev_dbg(dev, "%s: removing groups %p\n", __func__, groups);
sysfs_remove_groups(&dev->kobj, groups);
}
/**
* devm_device_add_group - given a device, create a managed attribute group
* @dev: The device to create the group for
* @grp: The attribute group to create
*
* This function creates a group for the first time. It will explicitly
* warn and error if any of the attribute files being created already exist.
*
* Returns 0 on success or error code on failure.
*/
int devm_device_add_group(struct device *dev, const struct attribute_group *grp)
{
union device_attr_group_devres *devres;
int error;
devres = devres_alloc(devm_attr_group_remove,
sizeof(*devres), GFP_KERNEL);
if (!devres)
return -ENOMEM;
error = sysfs_create_group(&dev->kobj, grp);
if (error) {
devres_free(devres);
return error;
}
devres->group = grp;
devres_add(dev, devres);
return 0;
}
EXPORT_SYMBOL_GPL(devm_device_add_group);
/**
* devm_device_remove_group: remove a managed group from a device
* @dev: device to remove the group from
* @grp: group to remove
*
* This function removes a group of attributes from a device. The attributes
* previously have to have been created for this group, otherwise it will fail.
*/
void devm_device_remove_group(struct device *dev,
const struct attribute_group *grp)
{
WARN_ON(devres_release(dev, devm_attr_group_remove,
devm_attr_group_match,
/* cast away const */ (void *)grp));
}
EXPORT_SYMBOL_GPL(devm_device_remove_group);
/**
* devm_device_add_groups - create a bunch of managed attribute groups
* @dev: The device to create the group for
* @groups: The attribute groups to create, NULL terminated
*
* This function creates a bunch of managed attribute groups. If an error
* occurs when creating a group, all previously created groups will be
* removed, unwinding everything back to the original state when this
* function was called. It will explicitly warn and error if any of the
* attribute files being created already exist.
*
* Returns 0 on success or error code from sysfs_create_group on failure.
*/
int devm_device_add_groups(struct device *dev,
const struct attribute_group **groups)
{
union device_attr_group_devres *devres;
int error;
devres = devres_alloc(devm_attr_groups_remove,
sizeof(*devres), GFP_KERNEL);
if (!devres)
return -ENOMEM;
error = sysfs_create_groups(&dev->kobj, groups);
if (error) {
devres_free(devres);
return error;
}
devres->groups = groups;
devres_add(dev, devres);
return 0;
}
EXPORT_SYMBOL_GPL(devm_device_add_groups);
/**
* devm_device_remove_groups - remove a list of managed groups
*
* @dev: The device for the groups to be removed from
* @groups: NULL terminated list of groups to be removed
*
* If groups is not NULL, remove the specified groups from the device.
*/
void devm_device_remove_groups(struct device *dev,
const struct attribute_group **groups)
{
WARN_ON(devres_release(dev, devm_attr_groups_remove,
devm_attr_group_match,
/* cast away const */ (void *)groups));
}
EXPORT_SYMBOL_GPL(devm_device_remove_groups);
static int device_add_attrs(struct device *dev)
{
struct class *class = dev->class;
const struct device_type *type = dev->type;
int error;
if (class) {
error = device_add_groups(dev, class->dev_groups);
if (error)
return error;
}
if (type) {
error = device_add_groups(dev, type->groups);
if (error)
goto err_remove_class_groups;
}
error = device_add_groups(dev, dev->groups);
if (error)
goto err_remove_type_groups;
Driver core: Add offline/online device operations In some cases, graceful hot-removal of devices is not possible, although in principle the devices in question support hotplug. For example, that may happen for the last CPU in the system or for memory modules holding kernel memory. In those cases it is nice to be able to check if the given device can be gracefully hot-removed before triggering a removal procedure that cannot be aborted or reversed. Unfortunately, however, the kernel currently doesn't provide any support for that. To address that deficiency, introduce support for offline and online operations that can be performed on devices, respectively, before a hot-removal and in case when it is necessary (or convenient) to put a device back online after a successful offline (that has not been followed by removal). The idea is that the offline will fail whenever the given device cannot be gracefully removed from the system and it will not be allowed to use the device after a successful offline (until a subsequent online) in analogy with the existing CPU offline/online mechanism. For now, the offline and online operations are introduced at the bus type level, as that should be sufficient for the most urgent use cases (CPUs and memory modules). In the future, however, the approach may be extended to cover some more complicated device offline/online scenarios involving device drivers etc. The lock_device_hotplug() and unlock_device_hotplug() functions are introduced because subsequent patches need to put larger pieces of code under device_hotplug_lock to prevent race conditions between device offline and removal from happening. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Reviewed-by: Toshi Kani <toshi.kani@hp.com>
2013-05-02 20:15:29 +00:00
if (device_supports_offline(dev) && !dev->offline_disabled) {
error = device_create_file(dev, &dev_attr_online);
Driver core: Add offline/online device operations In some cases, graceful hot-removal of devices is not possible, although in principle the devices in question support hotplug. For example, that may happen for the last CPU in the system or for memory modules holding kernel memory. In those cases it is nice to be able to check if the given device can be gracefully hot-removed before triggering a removal procedure that cannot be aborted or reversed. Unfortunately, however, the kernel currently doesn't provide any support for that. To address that deficiency, introduce support for offline and online operations that can be performed on devices, respectively, before a hot-removal and in case when it is necessary (or convenient) to put a device back online after a successful offline (that has not been followed by removal). The idea is that the offline will fail whenever the given device cannot be gracefully removed from the system and it will not be allowed to use the device after a successful offline (until a subsequent online) in analogy with the existing CPU offline/online mechanism. For now, the offline and online operations are introduced at the bus type level, as that should be sufficient for the most urgent use cases (CPUs and memory modules). In the future, however, the approach may be extended to cover some more complicated device offline/online scenarios involving device drivers etc. The lock_device_hotplug() and unlock_device_hotplug() functions are introduced because subsequent patches need to put larger pieces of code under device_hotplug_lock to prevent race conditions between device offline and removal from happening. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Reviewed-by: Toshi Kani <toshi.kani@hp.com>
2013-05-02 20:15:29 +00:00
if (error)
goto err_remove_dev_groups;
Driver core: Add offline/online device operations In some cases, graceful hot-removal of devices is not possible, although in principle the devices in question support hotplug. For example, that may happen for the last CPU in the system or for memory modules holding kernel memory. In those cases it is nice to be able to check if the given device can be gracefully hot-removed before triggering a removal procedure that cannot be aborted or reversed. Unfortunately, however, the kernel currently doesn't provide any support for that. To address that deficiency, introduce support for offline and online operations that can be performed on devices, respectively, before a hot-removal and in case when it is necessary (or convenient) to put a device back online after a successful offline (that has not been followed by removal). The idea is that the offline will fail whenever the given device cannot be gracefully removed from the system and it will not be allowed to use the device after a successful offline (until a subsequent online) in analogy with the existing CPU offline/online mechanism. For now, the offline and online operations are introduced at the bus type level, as that should be sufficient for the most urgent use cases (CPUs and memory modules). In the future, however, the approach may be extended to cover some more complicated device offline/online scenarios involving device drivers etc. The lock_device_hotplug() and unlock_device_hotplug() functions are introduced because subsequent patches need to put larger pieces of code under device_hotplug_lock to prevent race conditions between device offline and removal from happening. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Reviewed-by: Toshi Kani <toshi.kani@hp.com>
2013-05-02 20:15:29 +00:00
}
if (fw_devlink_flags && !fw_devlink_is_permissive() && dev->fwnode) {
error = device_create_file(dev, &dev_attr_waiting_for_supplier);
if (error)
goto err_remove_dev_online;
}
driver core: Move the "removable" attribute from USB to core Move the "removable" attribute from USB to core in order to allow it to be supported by other subsystem / buses. Individual buses that want to support this attribute can populate the removable property of the device while enumerating it with the 3 possible values - - "unknown" - "fixed" - "removable" Leaving the field unchanged (i.e. "not supported") would mean that the attribute would not show up in sysfs for that device. The UAPI (location, symantics etc) for the attribute remains unchanged. Move the "removable" attribute from USB to the device core so it can be used by other subsystems / buses. By default, devices do not have a "removable" attribute in sysfs. If a subsystem or bus driver wants to support a "removable" attribute, it should call device_set_removable() before calling device_register() or device_add(), e.g.: device_set_removable(dev, DEVICE_REMOVABLE); device_register(dev); The possible values and the resulting sysfs attribute contents are: DEVICE_REMOVABLE_UNKNOWN -> "unknown" DEVICE_REMOVABLE -> "removable" DEVICE_FIXED -> "fixed" Convert the USB "removable" attribute to use this new device core functionality. There should be no user-visible change in the location or semantics of attribute for USB devices. Reviewed-by: Bjorn Helgaas <bhelgaas@google.com> Signed-off-by: Rajat Jain <rajatja@google.com> Link: https://lore.kernel.org/r/20210524171812.18095-1-rajatja@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2021-05-24 17:18:11 +00:00
if (dev_removable_is_valid(dev)) {
error = device_create_file(dev, &dev_attr_removable);
if (error)
goto err_remove_dev_waiting_for_supplier;
}
if (dev_add_physical_location(dev)) {
error = device_add_group(dev,
&dev_attr_physical_location_group);
if (error)
goto err_remove_dev_removable;
}
return 0;
err_remove_dev_removable:
device_remove_file(dev, &dev_attr_removable);
driver core: Move the "removable" attribute from USB to core Move the "removable" attribute from USB to core in order to allow it to be supported by other subsystem / buses. Individual buses that want to support this attribute can populate the removable property of the device while enumerating it with the 3 possible values - - "unknown" - "fixed" - "removable" Leaving the field unchanged (i.e. "not supported") would mean that the attribute would not show up in sysfs for that device. The UAPI (location, symantics etc) for the attribute remains unchanged. Move the "removable" attribute from USB to the device core so it can be used by other subsystems / buses. By default, devices do not have a "removable" attribute in sysfs. If a subsystem or bus driver wants to support a "removable" attribute, it should call device_set_removable() before calling device_register() or device_add(), e.g.: device_set_removable(dev, DEVICE_REMOVABLE); device_register(dev); The possible values and the resulting sysfs attribute contents are: DEVICE_REMOVABLE_UNKNOWN -> "unknown" DEVICE_REMOVABLE -> "removable" DEVICE_FIXED -> "fixed" Convert the USB "removable" attribute to use this new device core functionality. There should be no user-visible change in the location or semantics of attribute for USB devices. Reviewed-by: Bjorn Helgaas <bhelgaas@google.com> Signed-off-by: Rajat Jain <rajatja@google.com> Link: https://lore.kernel.org/r/20210524171812.18095-1-rajatja@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2021-05-24 17:18:11 +00:00
err_remove_dev_waiting_for_supplier:
device_remove_file(dev, &dev_attr_waiting_for_supplier);
err_remove_dev_online:
device_remove_file(dev, &dev_attr_online);
err_remove_dev_groups:
device_remove_groups(dev, dev->groups);
err_remove_type_groups:
if (type)
device_remove_groups(dev, type->groups);
err_remove_class_groups:
if (class)
device_remove_groups(dev, class->dev_groups);
return error;
}
static void device_remove_attrs(struct device *dev)
{
struct class *class = dev->class;
const struct device_type *type = dev->type;
if (dev->physical_location) {
device_remove_group(dev, &dev_attr_physical_location_group);
kfree(dev->physical_location);
}
driver core: Move the "removable" attribute from USB to core Move the "removable" attribute from USB to core in order to allow it to be supported by other subsystem / buses. Individual buses that want to support this attribute can populate the removable property of the device while enumerating it with the 3 possible values - - "unknown" - "fixed" - "removable" Leaving the field unchanged (i.e. "not supported") would mean that the attribute would not show up in sysfs for that device. The UAPI (location, symantics etc) for the attribute remains unchanged. Move the "removable" attribute from USB to the device core so it can be used by other subsystems / buses. By default, devices do not have a "removable" attribute in sysfs. If a subsystem or bus driver wants to support a "removable" attribute, it should call device_set_removable() before calling device_register() or device_add(), e.g.: device_set_removable(dev, DEVICE_REMOVABLE); device_register(dev); The possible values and the resulting sysfs attribute contents are: DEVICE_REMOVABLE_UNKNOWN -> "unknown" DEVICE_REMOVABLE -> "removable" DEVICE_FIXED -> "fixed" Convert the USB "removable" attribute to use this new device core functionality. There should be no user-visible change in the location or semantics of attribute for USB devices. Reviewed-by: Bjorn Helgaas <bhelgaas@google.com> Signed-off-by: Rajat Jain <rajatja@google.com> Link: https://lore.kernel.org/r/20210524171812.18095-1-rajatja@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2021-05-24 17:18:11 +00:00
device_remove_file(dev, &dev_attr_removable);
device_remove_file(dev, &dev_attr_waiting_for_supplier);
device_remove_file(dev, &dev_attr_online);
device_remove_groups(dev, dev->groups);
if (type)
device_remove_groups(dev, type->groups);
if (class)
device_remove_groups(dev, class->dev_groups);
}
static ssize_t dev_show(struct device *dev, struct device_attribute *attr,
char *buf)
{
return print_dev_t(buf, dev->devt);
}
static DEVICE_ATTR_RO(dev);
/* /sys/devices/ */
struct kset *devices_kset;
driver core: correct device's shutdown order Now device's shutdown sequence is performed in reverse order of their registration in devices_kset list and this sequence corresponds to the reverse device's creation order. So, devices_kset data tracks "parent<-child" device's dependencies only. Unfortunately, that's not enough and causes problems in case of implementing board's specific shutdown procedures. For example [1]: "DRA7XX_evm uses PCF8575 and one of the PCF output lines feeds to MMC/SD and this line should be driven high in order for the MMC/SD to be detected. This line is modelled as regulator and the hsmmc driver takes care of enabling and disabling it. In the case of 'reboot', during shutdown path as part of it's cleanup process the hsmmc driver disables this regulator. This makes MMC boot not functional." To handle this issue the .shutdown() callback could be implemented for PCF8575 device where corresponding GPIO pins will be configured to states, required for correct warm/cold reset. This can be achieved only when all .shutdown() callbacks have been called already for all PCF8575's consumers. But devices_kset is not filled correctly now: devices_kset: Device61 4e000000.dmm devices_kset: Device62 48070000.i2c devices_kset: Device63 48072000.i2c devices_kset: Device64 48060000.i2c devices_kset: Device65 4809c000.mmc ... devices_kset: Device102 fixedregulator-sd ... devices_kset: Device181 0-0020 // PCF8575 devices_kset: Device182 gpiochip496 devices_kset: Device183 0-0021 // PCF8575 devices_kset: Device184 gpiochip480 As can be seen from above .shutdown() callback for PCF8575 will be called before its consumers, which, in turn means, that any changes of PCF8575 GPIO's pins will be or unsafe or overwritten later by GPIO's consumers. The problem can be solved if devices_kset list will be filled not only according device creation order, but also according device's probing order to track "supplier<-consumer" dependencies also. Hence, as a fix, lets add devices_kset_move_last(), devices_kset_move_before(), devices_kset_move_after() and call them from device_move() and also add call of devices_kset_move_last() in really_probe(). After this change all entries in devices_kset will be sorted according to device's creation ("parent<-child") and probing ("supplier<-consumer") order. devices_kset after: devices_kset: Device121 48070000.i2c devices_kset: Device122 i2c-0 ... devices_kset: Device147 regulator.24 devices_kset: Device148 0-0020 devices_kset: Device149 gpiochip496 devices_kset: Device150 0-0021 devices_kset: Device151 gpiochip480 devices_kset: Device152 0-0019 ... devices_kset: Device372 fixedregulator-sd devices_kset: Device373 regulator.29 devices_kset: Device374 4809c000.mmc devices_kset: Device375 mmc0 [1] http://www.spinics.net/lists/linux-mmc/msg29825.html Cc: Sekhar Nori <nsekhar@ti.com> Signed-off-by: Grygorii Strashko <grygorii.strashko@ti.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2015-07-27 17:43:01 +00:00
/**
* devices_kset_move_before - Move device in the devices_kset's list.
* @deva: Device to move.
* @devb: Device @deva should come before.
*/
static void devices_kset_move_before(struct device *deva, struct device *devb)
{
if (!devices_kset)
return;
pr_debug("devices_kset: Moving %s before %s\n",
dev_name(deva), dev_name(devb));
spin_lock(&devices_kset->list_lock);
list_move_tail(&deva->kobj.entry, &devb->kobj.entry);
spin_unlock(&devices_kset->list_lock);
}
/**
* devices_kset_move_after - Move device in the devices_kset's list.
* @deva: Device to move
* @devb: Device @deva should come after.
*/
static void devices_kset_move_after(struct device *deva, struct device *devb)
{
if (!devices_kset)
return;
pr_debug("devices_kset: Moving %s after %s\n",
dev_name(deva), dev_name(devb));
spin_lock(&devices_kset->list_lock);
list_move(&deva->kobj.entry, &devb->kobj.entry);
spin_unlock(&devices_kset->list_lock);
}
/**
* devices_kset_move_last - move the device to the end of devices_kset's list.
* @dev: device to move
*/
void devices_kset_move_last(struct device *dev)
{
if (!devices_kset)
return;
pr_debug("devices_kset: Moving %s to end of list\n", dev_name(dev));
spin_lock(&devices_kset->list_lock);
list_move_tail(&dev->kobj.entry, &devices_kset->list);
spin_unlock(&devices_kset->list_lock);
}
/**
* device_create_file - create sysfs attribute file for device.
* @dev: device.
* @attr: device attribute descriptor.
*/
int device_create_file(struct device *dev,
const struct device_attribute *attr)
{
int error = 0;
if (dev) {
WARN(((attr->attr.mode & S_IWUGO) && !attr->store),
"Attribute %s: write permission without 'store'\n",
attr->attr.name);
WARN(((attr->attr.mode & S_IRUGO) && !attr->show),
"Attribute %s: read permission without 'show'\n",
attr->attr.name);
error = sysfs_create_file(&dev->kobj, &attr->attr);
}
return error;
}
EXPORT_SYMBOL_GPL(device_create_file);
/**
* device_remove_file - remove sysfs attribute file.
* @dev: device.
* @attr: device attribute descriptor.
*/
void device_remove_file(struct device *dev,
const struct device_attribute *attr)
{
if (dev)
sysfs_remove_file(&dev->kobj, &attr->attr);
}
EXPORT_SYMBOL_GPL(device_remove_file);
kernfs, sysfs, driver-core: implement kernfs_remove_self() and its wrappers Sometimes it's necessary to implement a node which wants to delete nodes including itself. This isn't straightforward because of kernfs active reference. While a file operation is in progress, an active reference is held and kernfs_remove() waits for all such references to drain before completing. For a self-deleting node, this is a deadlock as kernfs_remove() ends up waiting for an active reference that itself is sitting on top of. This currently is worked around in the sysfs layer using sysfs_schedule_callback() which makes such removals asynchronous. While it works, it's rather cumbersome and inherently breaks synchronicity of the operation - the file operation which triggered the operation may complete before the removal is finished (or even started) and the removal may fail asynchronously. If a removal operation is immmediately followed by another operation which expects the specific name to be available (e.g. removal followed by rename onto the same name), there's no way to make the latter operation reliable. The thing is there's no inherent reason for this to be asynchrnous. All that's necessary to do this synchronous is a dedicated operation which drops its own active ref and deactivates self. This patch implements kernfs_remove_self() and its wrappers in sysfs and driver core. kernfs_remove_self() is to be called from one of the file operations, drops the active ref the task is holding, removes the self node, and restores active ref to the dead node so that the ref is balanced afterwards. __kernfs_remove() is updated so that it takes an early exit if the target node is already fully removed so that the active ref restored by kernfs_remove_self() after removal doesn't confuse the deactivation path. This makes implementing self-deleting nodes very easy. The normal removal path doesn't even need to be changed to use kernfs_remove_self() for the self-deleting node. The method can invoke kernfs_remove_self() on itself before proceeding the normal removal path. kernfs_remove() invoked on the node by the normal deletion path will simply be ignored. This will replace sysfs_schedule_callback(). A subtle feature of sysfs_schedule_callback() is that it collapses multiple invocations - even if multiple removals are triggered, the removal callback is run only once. An equivalent effect can be achieved by testing the return value of kernfs_remove_self() - only the one which gets %true return value should proceed with actual deletion. All other instances of kernfs_remove_self() will wait till the enclosing kernfs operation which invoked the winning instance of kernfs_remove_self() finishes and then return %false. This trivially makes all users of kernfs_remove_self() automatically show correct synchronous behavior even when there are multiple concurrent operations - all "echo 1 > delete" instances will finish only after the whole operation is completed by one of the instances. Note that manipulation of active ref is implemented in separate public functions - kernfs_[un]break_active_protection(). kernfs_remove_self() is the only user at the moment but this will be used to cater to more complex cases. v2: For !CONFIG_SYSFS, dummy version kernfs_remove_self() was missing and sysfs_remove_file_self() had incorrect return type. Fix it. Reported by kbuild test bot. v3: kernfs_[un]break_active_protection() separated out from kernfs_remove_self() and exposed as public API. Signed-off-by: Tejun Heo <tj@kernel.org> Cc: Alan Stern <stern@rowland.harvard.edu> Cc: kbuild test robot <fengguang.wu@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2014-02-03 19:03:01 +00:00
/**
* device_remove_file_self - remove sysfs attribute file from its own method.
* @dev: device.
* @attr: device attribute descriptor.
*
* See kernfs_remove_self() for details.
*/
bool device_remove_file_self(struct device *dev,
const struct device_attribute *attr)
{
if (dev)
return sysfs_remove_file_self(&dev->kobj, &attr->attr);
else
return false;
}
EXPORT_SYMBOL_GPL(device_remove_file_self);
/**
* device_create_bin_file - create sysfs binary attribute file for device.
* @dev: device.
* @attr: device binary attribute descriptor.
*/
int device_create_bin_file(struct device *dev,
const struct bin_attribute *attr)
{
int error = -EINVAL;
if (dev)
error = sysfs_create_bin_file(&dev->kobj, attr);
return error;
}
EXPORT_SYMBOL_GPL(device_create_bin_file);
/**
* device_remove_bin_file - remove sysfs binary attribute file
* @dev: device.
* @attr: device binary attribute descriptor.
*/
void device_remove_bin_file(struct device *dev,
const struct bin_attribute *attr)
{
if (dev)
sysfs_remove_bin_file(&dev->kobj, attr);
}
EXPORT_SYMBOL_GPL(device_remove_bin_file);
static void klist_children_get(struct klist_node *n)
{
struct device_private *p = to_device_private_parent(n);
struct device *dev = p->device;
get_device(dev);
}
static void klist_children_put(struct klist_node *n)
{
struct device_private *p = to_device_private_parent(n);
struct device *dev = p->device;
put_device(dev);
}
/**
* device_initialize - init device structure.
* @dev: device.
*
* This prepares the device for use by other layers by initializing
* its fields.
* It is the first half of device_register(), if called by
* that function, though it can also be called separately, so one
* may use @dev's fields. In particular, get_device()/put_device()
* may be used for reference counting of @dev after calling this
* function.
*
* All fields in @dev must be initialized by the caller to 0, except
* for those explicitly set to some other value. The simplest
* approach is to use kzalloc() to allocate the structure containing
* @dev.
*
* NOTE: Use put_device() to give up your reference instead of freeing
* @dev directly once you have called this function.
*/
void device_initialize(struct device *dev)
{
dev->kobj.kset = devices_kset;
kobject_init(&dev->kobj, &device_ktype);
INIT_LIST_HEAD(&dev->dma_pools);
mutex_init(&dev->mutex);
lockdep_set_novalidate_class(&dev->mutex);
spin_lock_init(&dev->devres_lock);
INIT_LIST_HEAD(&dev->devres_head);
device_pm_init(dev);
set_dev_node(dev, NUMA_NO_NODE);
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
INIT_LIST_HEAD(&dev->links.consumers);
INIT_LIST_HEAD(&dev->links.suppliers);
INIT_LIST_HEAD(&dev->links.defer_sync);
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
dev->links.status = DL_DEV_NO_DRIVER;
#if defined(CONFIG_ARCH_HAS_SYNC_DMA_FOR_DEVICE) || \
defined(CONFIG_ARCH_HAS_SYNC_DMA_FOR_CPU) || \
defined(CONFIG_ARCH_HAS_SYNC_DMA_FOR_CPU_ALL)
dev->dma_coherent = dma_default_coherent;
#endif
#ifdef CONFIG_SWIOTLB
swiotlb: Convert io_default_tlb_mem to static allocation Since commit 69031f500865 ("swiotlb: Set dev->dma_io_tlb_mem to the swiotlb pool used"), 'struct device' may hold a copy of the global 'io_default_tlb_mem' pointer if the device is using swiotlb for DMA. A subsequent call to swiotlb_exit() will therefore leave dangling pointers behind in these device structures, resulting in KASAN splats such as: | BUG: KASAN: use-after-free in __iommu_dma_unmap_swiotlb+0x64/0xb0 | Read of size 8 at addr ffff8881d7830000 by task swapper/0/0 | | CPU: 0 PID: 0 Comm: swapper/0 Not tainted 5.12.0-rc3-debug #1 | Hardware name: HP HP Desktop M01-F1xxx/87D6, BIOS F.12 12/17/2020 | Call Trace: | <IRQ> | dump_stack+0x9c/0xcf | print_address_description.constprop.0+0x18/0x130 | kasan_report.cold+0x7f/0x111 | __iommu_dma_unmap_swiotlb+0x64/0xb0 | nvme_pci_complete_rq+0x73/0x130 | blk_complete_reqs+0x6f/0x80 | __do_softirq+0xfc/0x3be Convert 'io_default_tlb_mem' to a static structure, so that the per-device pointers remain valid after swiotlb_exit() has been invoked. All users are updated to reference the static structure directly, using the 'nslabs' field to determine whether swiotlb has been initialised. The 'slots' array is still allocated dynamically and referenced via a pointer rather than a flexible array member. Cc: Claire Chang <tientzu@chromium.org> Cc: Christoph Hellwig <hch@lst.de> Cc: Robin Murphy <robin.murphy@arm.com> Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> Fixes: 69031f500865 ("swiotlb: Set dev->dma_io_tlb_mem to the swiotlb pool used") Reported-by: Nathan Chancellor <nathan@kernel.org> Tested-by: Nathan Chancellor <nathan@kernel.org> Tested-by: Claire Chang <tientzu@chromium.org> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Will Deacon <will@kernel.org> Signed-off-by: Konrad Rzeszutek Wilk <konrad@kernel.org>
2021-07-20 13:38:24 +00:00
dev->dma_io_tlb_mem = &io_tlb_default_mem;
#endif
}
EXPORT_SYMBOL_GPL(device_initialize);
struct kobject *virtual_device_parent(struct device *dev)
{
static struct kobject *virtual_dir = NULL;
if (!virtual_dir)
virtual_dir = kobject_create_and_add("virtual",
&devices_kset->kobj);
return virtual_dir;
}
struct class_dir {
struct kobject kobj;
struct class *class;
};
#define to_class_dir(obj) container_of(obj, struct class_dir, kobj)
static void class_dir_release(struct kobject *kobj)
{
struct class_dir *dir = to_class_dir(kobj);
kfree(dir);
}
static const
struct kobj_ns_type_operations *class_dir_child_ns_type(struct kobject *kobj)
{
struct class_dir *dir = to_class_dir(kobj);
return dir->class->ns_type;
}
static struct kobj_type class_dir_ktype = {
.release = class_dir_release,
.sysfs_ops = &kobj_sysfs_ops,
.child_ns_type = class_dir_child_ns_type
};
static struct kobject *
class_dir_create_and_add(struct class *class, struct kobject *parent_kobj)
{
struct class_dir *dir;
int retval;
dir = kzalloc(sizeof(*dir), GFP_KERNEL);
if (!dir)
return ERR_PTR(-ENOMEM);
dir->class = class;
kobject_init(&dir->kobj, &class_dir_ktype);
dir->kobj.kset = &class->p->glue_dirs;
retval = kobject_add(&dir->kobj, parent_kobj, "%s", class->name);
if (retval < 0) {
kobject_put(&dir->kobj);
return ERR_PTR(retval);
}
return &dir->kobj;
}
sysfs: driver core: Fix glue dir race condition by gdp_mutex There is a race condition when removing glue directory. It can be reproduced in following test: path 1: Add first child device device_add() get_device_parent() /*find parent from glue_dirs.list*/ list_for_each_entry(k, &dev->class->p->glue_dirs.list, entry) if (k->parent == parent_kobj) { kobj = kobject_get(k); break; } .... class_dir_create_and_add() path2: Remove last child device under glue dir device_del() cleanup_device_parent() cleanup_glue_dir() kobject_put(glue_dir); If path2 has been called cleanup_glue_dir(), but not call kobject_put(glue_dir), the glue dir is still in parent's kset list. Meanwhile, path1 find the glue dir from the glue_dirs.list. Path2 may release glue dir before path1 call kobject_get(). So kernel will report the warning and bug_on. This is a "classic" problem we have of a kref in a list that can be found while the last instance could be removed at the same time. This patch reuse gdp_mutex to fix this race condition. The following calltrace is captured in kernel 3.4, but the latest kernel still has this bug. ----------------------------------------------------- <4>[ 3965.441471] WARNING: at ...include/linux/kref.h:41 kobject_get+0x33/0x40() <4>[ 3965.441474] Hardware name: Romley <4>[ 3965.441475] Modules linked in: isd_iop(O) isd_xda(O)... ... <4>[ 3965.441605] Call Trace: <4>[ 3965.441611] [<ffffffff8103717a>] warn_slowpath_common+0x7a/0xb0 <4>[ 3965.441615] [<ffffffff810371c5>] warn_slowpath_null+0x15/0x20 <4>[ 3965.441618] [<ffffffff81215963>] kobject_get+0x33/0x40 <4>[ 3965.441624] [<ffffffff812d1e45>] get_device_parent.isra.11+0x135/0x1f0 <4>[ 3965.441627] [<ffffffff812d22d4>] device_add+0xd4/0x6d0 <4>[ 3965.441631] [<ffffffff812d0dbc>] ? dev_set_name+0x3c/0x40 .... <2>[ 3965.441912] kernel BUG at ..../fs/sysfs/group.c:65! <4>[ 3965.441915] invalid opcode: 0000 [#1] SMP ... <4>[ 3965.686743] [<ffffffff811a677e>] sysfs_create_group+0xe/0x10 <4>[ 3965.686748] [<ffffffff810cfb04>] blk_trace_init_sysfs+0x14/0x20 <4>[ 3965.686753] [<ffffffff811fcabb>] blk_register_queue+0x3b/0x120 <4>[ 3965.686756] [<ffffffff812030bc>] add_disk+0x1cc/0x490 .... ------------------------------------------------------- Signed-off-by: Yijing Wang <wangyijing@huawei.com> Signed-off-by: Weng Meiling <wengmeiling.weng@huawei.com> Cc: <stable@vger.kernel.org> #3.4+ Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2014-11-07 04:05:49 +00:00
static DEFINE_MUTEX(gdp_mutex);
static struct kobject *get_device_parent(struct device *dev,
struct device *parent)
{
if (dev->class) {
struct kobject *kobj = NULL;
struct kobject *parent_kobj;
struct kobject *k;
#ifdef CONFIG_BLOCK
/* block disks show up in /sys/block */
if (sysfs_deprecated && dev->class == &block_class) {
if (parent && parent->class == &block_class)
return &parent->kobj;
return &block_class.p->subsys.kobj;
}
#endif
/*
* If we have no parent, we live in "virtual".
* Class-devices with a non class-device as parent, live
* in a "glue" directory to prevent namespace collisions.
*/
if (parent == NULL)
parent_kobj = virtual_device_parent(dev);
Driver-core: Always create class directories for classses that support namespaces. This fixes the regression in 2.6.35-rcX where bluetooth network devices would fail to be deleted from sysfs, causing their destruction and recreation to fail. In addition this fixes the mac80211_hwsim driver where it would leave around sysfs files when the driver was removed. This problem is discussed at https://bugzilla.kernel.org/show_bug.cgi?id=16257 The reason for the regression is that the network namespace support added to sysfs expects and requires that network devices be put in directories that can contain only network devices. Today get_device_parent almost provides that guarantee for all class devices, except for a specific exception when the parent of a class devices is a class device. It would be nice to simply remove that arguably incorrect special case, but apparently the input devices depend on it being there. So I have only removed it for class devices with network namespace support. Which today are the network devices. It has been suggested that a better fix would be to change the parent device from a class device to a bus device, which in the case of the bluetooth driver would change /sys/class/bluetooth to /sys/bus/bluetoth, I can not see how we would avoid significant userspace breakage if we were to make that change. Adding an extra directory in the path to the device will also be userspace visible but it is much less likely to break things. Everything is still accessible from /sys/class (for example), and it fixes two bugs. Adding an extra directory fixes a 3 year old regression introduced with the new sysfs layout that makes it impossible to rename bnep0 network devices to names that conflict with hci device attributes like hci_revsion. Adding an additional directory removes the new failure modes introduced by the network namespace code. If it weren't for the regession in the renaming of network devices I would figure out how to just make the sysfs code deal with this configuration of devices. In summary this patch fixes regressions by changing: "/sys/class/bluetooth/hci0/bnep0" to "/sys/class/bluetooth/hci0/net/bnep0". Reported-by: Johannes Berg <johannes@sipsolutions.net> Reported-by: Janusz Krzysztofik <jkrzyszt@tis.icnet.pl> Signed-off-by: Eric W. Biederman <ebiederm@xmission.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2010-07-25 05:43:35 +00:00
else if (parent->class && !dev->class->ns_type)
return &parent->kobj;
else
parent_kobj = &parent->kobj;
driver-core: fix race condition in get_device_parent() sysfs is creating several devices in cuse class concurrently and with CONFIG_SYSFS_DEPRECATED turned off, it triggers the following oops. BUG: unable to handle kernel NULL pointer dereference at 0000000000000038 IP: [<ffffffff81158b0a>] sysfs_addrm_start+0x4a/0xf0 PGD 75bb067 PUD 75be067 PMD 0 Oops: 0000 [#1] PREEMPT SMP last sysfs file: /sys/devices/system/cpu/cpu7/topology/core_siblings CPU 1 Modules linked in: cuse fuse Pid: 4737, comm: osspd Not tainted 2.6.31-work #77 RIP: 0010:[<ffffffff81158b0a>] [<ffffffff81158b0a>] sysfs_addrm_start+0x4a/0xf0 RSP: 0018:ffff88000042f8f8 EFLAGS: 00010296 RAX: ffff88000042ffd8 RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffff880007eef660 RDI: 0000000000000001 RBP: ffff88000042f918 R08: 0000000000000000 R09: 0000000000000000 R10: 0000000000000001 R11: ffffffff81158b0a R12: ffff88000042f928 R13: 00000000fffffff4 R14: 0000000000000000 R15: ffff88000042f9a0 FS: 00007fe93905a950(0000) GS:ffff880008600000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b CR2: 0000000000000038 CR3: 00000000077c9000 CR4: 00000000000006e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 Process osspd (pid: 4737, threadinfo ffff88000042e000, task ffff880007eef040) Stack: ffff880005da10e8 0000000011cc8d6e ffff88000042f928 ffff880003d28a28 <0> ffff88000042f988 ffffffff811592d7 0000000000000000 0000000000000000 <0> 0000000000000000 0000000000000000 ffff88000042f958 0000000011cc8d6e Call Trace: [<ffffffff811592d7>] create_dir+0x67/0xe0 [<ffffffff811593a8>] sysfs_create_dir+0x58/0xb0 [<ffffffff8128ca7c>] ? kobject_add_internal+0xcc/0x220 [<ffffffff812942e1>] ? vsnprintf+0x3c1/0xb90 [<ffffffff8128cab7>] kobject_add_internal+0x107/0x220 [<ffffffff8128cd37>] kobject_add_varg+0x47/0x80 [<ffffffff8128ce53>] kobject_add+0x53/0x90 [<ffffffff81357d84>] device_add+0xd4/0x690 [<ffffffff81356c2b>] ? dev_set_name+0x4b/0x70 [<ffffffffa001a884>] cuse_process_init_reply+0x2b4/0x420 [cuse] ... The problem is that kobject_add_internal() first adds a kobject to the kset and then try to create sysfs directory for it. If the creation fails, it remove the kobject from the kset. get_device_parent() accesses class_dirs kset while only holding class_dirs.list_lock to see whether the cuse class dir exists. But when it exists, it may not have finished initialization yet or may fail and get removed soon. In the above case, the former happened so the second one ends up trying to create subdirectory under NULL sysfs_dirent. Fix it by grabbing a mutex in get_device_parent(). Signed-off-by: Tejun Heo <tj@kernel.org> Reported-by: Colin Guthrie <cguthrie@mandriva.org> Cc: stable <stable@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2010-02-05 08:57:02 +00:00
mutex_lock(&gdp_mutex);
/* find our class-directory at the parent and reference it */
spin_lock(&dev->class->p->glue_dirs.list_lock);
list_for_each_entry(k, &dev->class->p->glue_dirs.list, entry)
if (k->parent == parent_kobj) {
kobj = kobject_get(k);
break;
}
spin_unlock(&dev->class->p->glue_dirs.list_lock);
driver-core: fix race condition in get_device_parent() sysfs is creating several devices in cuse class concurrently and with CONFIG_SYSFS_DEPRECATED turned off, it triggers the following oops. BUG: unable to handle kernel NULL pointer dereference at 0000000000000038 IP: [<ffffffff81158b0a>] sysfs_addrm_start+0x4a/0xf0 PGD 75bb067 PUD 75be067 PMD 0 Oops: 0000 [#1] PREEMPT SMP last sysfs file: /sys/devices/system/cpu/cpu7/topology/core_siblings CPU 1 Modules linked in: cuse fuse Pid: 4737, comm: osspd Not tainted 2.6.31-work #77 RIP: 0010:[<ffffffff81158b0a>] [<ffffffff81158b0a>] sysfs_addrm_start+0x4a/0xf0 RSP: 0018:ffff88000042f8f8 EFLAGS: 00010296 RAX: ffff88000042ffd8 RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffff880007eef660 RDI: 0000000000000001 RBP: ffff88000042f918 R08: 0000000000000000 R09: 0000000000000000 R10: 0000000000000001 R11: ffffffff81158b0a R12: ffff88000042f928 R13: 00000000fffffff4 R14: 0000000000000000 R15: ffff88000042f9a0 FS: 00007fe93905a950(0000) GS:ffff880008600000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b CR2: 0000000000000038 CR3: 00000000077c9000 CR4: 00000000000006e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 Process osspd (pid: 4737, threadinfo ffff88000042e000, task ffff880007eef040) Stack: ffff880005da10e8 0000000011cc8d6e ffff88000042f928 ffff880003d28a28 <0> ffff88000042f988 ffffffff811592d7 0000000000000000 0000000000000000 <0> 0000000000000000 0000000000000000 ffff88000042f958 0000000011cc8d6e Call Trace: [<ffffffff811592d7>] create_dir+0x67/0xe0 [<ffffffff811593a8>] sysfs_create_dir+0x58/0xb0 [<ffffffff8128ca7c>] ? kobject_add_internal+0xcc/0x220 [<ffffffff812942e1>] ? vsnprintf+0x3c1/0xb90 [<ffffffff8128cab7>] kobject_add_internal+0x107/0x220 [<ffffffff8128cd37>] kobject_add_varg+0x47/0x80 [<ffffffff8128ce53>] kobject_add+0x53/0x90 [<ffffffff81357d84>] device_add+0xd4/0x690 [<ffffffff81356c2b>] ? dev_set_name+0x4b/0x70 [<ffffffffa001a884>] cuse_process_init_reply+0x2b4/0x420 [cuse] ... The problem is that kobject_add_internal() first adds a kobject to the kset and then try to create sysfs directory for it. If the creation fails, it remove the kobject from the kset. get_device_parent() accesses class_dirs kset while only holding class_dirs.list_lock to see whether the cuse class dir exists. But when it exists, it may not have finished initialization yet or may fail and get removed soon. In the above case, the former happened so the second one ends up trying to create subdirectory under NULL sysfs_dirent. Fix it by grabbing a mutex in get_device_parent(). Signed-off-by: Tejun Heo <tj@kernel.org> Reported-by: Colin Guthrie <cguthrie@mandriva.org> Cc: stable <stable@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2010-02-05 08:57:02 +00:00
if (kobj) {
mutex_unlock(&gdp_mutex);
return kobj;
driver-core: fix race condition in get_device_parent() sysfs is creating several devices in cuse class concurrently and with CONFIG_SYSFS_DEPRECATED turned off, it triggers the following oops. BUG: unable to handle kernel NULL pointer dereference at 0000000000000038 IP: [<ffffffff81158b0a>] sysfs_addrm_start+0x4a/0xf0 PGD 75bb067 PUD 75be067 PMD 0 Oops: 0000 [#1] PREEMPT SMP last sysfs file: /sys/devices/system/cpu/cpu7/topology/core_siblings CPU 1 Modules linked in: cuse fuse Pid: 4737, comm: osspd Not tainted 2.6.31-work #77 RIP: 0010:[<ffffffff81158b0a>] [<ffffffff81158b0a>] sysfs_addrm_start+0x4a/0xf0 RSP: 0018:ffff88000042f8f8 EFLAGS: 00010296 RAX: ffff88000042ffd8 RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffff880007eef660 RDI: 0000000000000001 RBP: ffff88000042f918 R08: 0000000000000000 R09: 0000000000000000 R10: 0000000000000001 R11: ffffffff81158b0a R12: ffff88000042f928 R13: 00000000fffffff4 R14: 0000000000000000 R15: ffff88000042f9a0 FS: 00007fe93905a950(0000) GS:ffff880008600000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b CR2: 0000000000000038 CR3: 00000000077c9000 CR4: 00000000000006e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 Process osspd (pid: 4737, threadinfo ffff88000042e000, task ffff880007eef040) Stack: ffff880005da10e8 0000000011cc8d6e ffff88000042f928 ffff880003d28a28 <0> ffff88000042f988 ffffffff811592d7 0000000000000000 0000000000000000 <0> 0000000000000000 0000000000000000 ffff88000042f958 0000000011cc8d6e Call Trace: [<ffffffff811592d7>] create_dir+0x67/0xe0 [<ffffffff811593a8>] sysfs_create_dir+0x58/0xb0 [<ffffffff8128ca7c>] ? kobject_add_internal+0xcc/0x220 [<ffffffff812942e1>] ? vsnprintf+0x3c1/0xb90 [<ffffffff8128cab7>] kobject_add_internal+0x107/0x220 [<ffffffff8128cd37>] kobject_add_varg+0x47/0x80 [<ffffffff8128ce53>] kobject_add+0x53/0x90 [<ffffffff81357d84>] device_add+0xd4/0x690 [<ffffffff81356c2b>] ? dev_set_name+0x4b/0x70 [<ffffffffa001a884>] cuse_process_init_reply+0x2b4/0x420 [cuse] ... The problem is that kobject_add_internal() first adds a kobject to the kset and then try to create sysfs directory for it. If the creation fails, it remove the kobject from the kset. get_device_parent() accesses class_dirs kset while only holding class_dirs.list_lock to see whether the cuse class dir exists. But when it exists, it may not have finished initialization yet or may fail and get removed soon. In the above case, the former happened so the second one ends up trying to create subdirectory under NULL sysfs_dirent. Fix it by grabbing a mutex in get_device_parent(). Signed-off-by: Tejun Heo <tj@kernel.org> Reported-by: Colin Guthrie <cguthrie@mandriva.org> Cc: stable <stable@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2010-02-05 08:57:02 +00:00
}
/* or create a new class-directory at the parent device */
k = class_dir_create_and_add(dev->class, parent_kobj);
/* do not emit an uevent for this simple "glue" directory */
driver-core: fix race condition in get_device_parent() sysfs is creating several devices in cuse class concurrently and with CONFIG_SYSFS_DEPRECATED turned off, it triggers the following oops. BUG: unable to handle kernel NULL pointer dereference at 0000000000000038 IP: [<ffffffff81158b0a>] sysfs_addrm_start+0x4a/0xf0 PGD 75bb067 PUD 75be067 PMD 0 Oops: 0000 [#1] PREEMPT SMP last sysfs file: /sys/devices/system/cpu/cpu7/topology/core_siblings CPU 1 Modules linked in: cuse fuse Pid: 4737, comm: osspd Not tainted 2.6.31-work #77 RIP: 0010:[<ffffffff81158b0a>] [<ffffffff81158b0a>] sysfs_addrm_start+0x4a/0xf0 RSP: 0018:ffff88000042f8f8 EFLAGS: 00010296 RAX: ffff88000042ffd8 RBX: 0000000000000000 RCX: 0000000000000000 RDX: 0000000000000000 RSI: ffff880007eef660 RDI: 0000000000000001 RBP: ffff88000042f918 R08: 0000000000000000 R09: 0000000000000000 R10: 0000000000000001 R11: ffffffff81158b0a R12: ffff88000042f928 R13: 00000000fffffff4 R14: 0000000000000000 R15: ffff88000042f9a0 FS: 00007fe93905a950(0000) GS:ffff880008600000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b CR2: 0000000000000038 CR3: 00000000077c9000 CR4: 00000000000006e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 Process osspd (pid: 4737, threadinfo ffff88000042e000, task ffff880007eef040) Stack: ffff880005da10e8 0000000011cc8d6e ffff88000042f928 ffff880003d28a28 <0> ffff88000042f988 ffffffff811592d7 0000000000000000 0000000000000000 <0> 0000000000000000 0000000000000000 ffff88000042f958 0000000011cc8d6e Call Trace: [<ffffffff811592d7>] create_dir+0x67/0xe0 [<ffffffff811593a8>] sysfs_create_dir+0x58/0xb0 [<ffffffff8128ca7c>] ? kobject_add_internal+0xcc/0x220 [<ffffffff812942e1>] ? vsnprintf+0x3c1/0xb90 [<ffffffff8128cab7>] kobject_add_internal+0x107/0x220 [<ffffffff8128cd37>] kobject_add_varg+0x47/0x80 [<ffffffff8128ce53>] kobject_add+0x53/0x90 [<ffffffff81357d84>] device_add+0xd4/0x690 [<ffffffff81356c2b>] ? dev_set_name+0x4b/0x70 [<ffffffffa001a884>] cuse_process_init_reply+0x2b4/0x420 [cuse] ... The problem is that kobject_add_internal() first adds a kobject to the kset and then try to create sysfs directory for it. If the creation fails, it remove the kobject from the kset. get_device_parent() accesses class_dirs kset while only holding class_dirs.list_lock to see whether the cuse class dir exists. But when it exists, it may not have finished initialization yet or may fail and get removed soon. In the above case, the former happened so the second one ends up trying to create subdirectory under NULL sysfs_dirent. Fix it by grabbing a mutex in get_device_parent(). Signed-off-by: Tejun Heo <tj@kernel.org> Reported-by: Colin Guthrie <cguthrie@mandriva.org> Cc: stable <stable@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2010-02-05 08:57:02 +00:00
mutex_unlock(&gdp_mutex);
return k;
}
/* subsystems can specify a default root directory for their devices */
if (!parent && dev->bus && dev->bus->dev_root)
return &dev->bus->dev_root->kobj;
if (parent)
return &parent->kobj;
return NULL;
}
static inline bool live_in_glue_dir(struct kobject *kobj,
struct device *dev)
{
if (!kobj || !dev->class ||
kobj->kset != &dev->class->p->glue_dirs)
return false;
return true;
}
static inline struct kobject *get_glue_dir(struct device *dev)
{
return dev->kobj.parent;
}
/**
* kobject_has_children - Returns whether a kobject has children.
* @kobj: the object to test
*
* This will return whether a kobject has other kobjects as children.
*
* It does NOT account for the presence of attribute files, only sub
* directories. It also assumes there is no concurrent addition or
* removal of such children, and thus relies on external locking.
*/
static inline bool kobject_has_children(struct kobject *kobj)
{
WARN_ON_ONCE(kref_read(&kobj->kref) == 0);
return kobj->sd && kobj->sd->dir.subdirs;
}
/*
* make sure cleaning up dir as the last step, we need to make
* sure .release handler of kobject is run with holding the
* global lock
*/
static void cleanup_glue_dir(struct device *dev, struct kobject *glue_dir)
{
driver core: Fix use-after-free and double free on glue directory There is a race condition between removing glue directory and adding a new device under the glue dir. It can be reproduced in following test: CPU1: CPU2: device_add() get_device_parent() class_dir_create_and_add() kobject_add_internal() create_dir() // create glue_dir device_add() get_device_parent() kobject_get() // get glue_dir device_del() cleanup_glue_dir() kobject_del(glue_dir) kobject_add() kobject_add_internal() create_dir() // in glue_dir sysfs_create_dir_ns() kernfs_create_dir_ns(sd) sysfs_remove_dir() // glue_dir->sd=NULL sysfs_put() // free glue_dir->sd // sd is freed kernfs_new_node(sd) kernfs_get(glue_dir) kernfs_add_one() kernfs_put() Before CPU1 remove last child device under glue dir, if CPU2 add a new device under glue dir, the glue_dir kobject reference count will be increase to 2 via kobject_get() in get_device_parent(). And CPU2 has been called kernfs_create_dir_ns(), but not call kernfs_new_node(). Meanwhile, CPU1 call sysfs_remove_dir() and sysfs_put(). This result in glue_dir->sd is freed and it's reference count will be 0. Then CPU2 call kernfs_get(glue_dir) will trigger a warning in kernfs_get() and increase it's reference count to 1. Because glue_dir->sd is freed by CPU1, the next call kernfs_add_one() by CPU2 will fail(This is also use-after-free) and call kernfs_put() to decrease reference count. Because the reference count is decremented to 0, it will also call kmem_cache_free() to free the glue_dir->sd again. This will result in double free. In order to avoid this happening, we also should make sure that kernfs_node for glue_dir is released in CPU1 only when refcount for glue_dir kobj is 1 to fix this race. The following calltrace is captured in kernel 4.14 with the following patch applied: commit 726e41097920 ("drivers: core: Remove glue dirs from sysfs earlier") -------------------------------------------------------------------------- [ 3.633703] WARNING: CPU: 4 PID: 513 at .../fs/kernfs/dir.c:494 Here is WARN_ON(!atomic_read(&kn->count) in kernfs_get(). .... [ 3.633986] Call trace: [ 3.633991] kernfs_create_dir_ns+0xa8/0xb0 [ 3.633994] sysfs_create_dir_ns+0x54/0xe8 [ 3.634001] kobject_add_internal+0x22c/0x3f0 [ 3.634005] kobject_add+0xe4/0x118 [ 3.634011] device_add+0x200/0x870 [ 3.634017] _request_firmware+0x958/0xc38 [ 3.634020] request_firmware_into_buf+0x4c/0x70 .... [ 3.634064] kernel BUG at .../mm/slub.c:294! Here is BUG_ON(object == fp) in set_freepointer(). .... [ 3.634346] Call trace: [ 3.634351] kmem_cache_free+0x504/0x6b8 [ 3.634355] kernfs_put+0x14c/0x1d8 [ 3.634359] kernfs_create_dir_ns+0x88/0xb0 [ 3.634362] sysfs_create_dir_ns+0x54/0xe8 [ 3.634366] kobject_add_internal+0x22c/0x3f0 [ 3.634370] kobject_add+0xe4/0x118 [ 3.634374] device_add+0x200/0x870 [ 3.634378] _request_firmware+0x958/0xc38 [ 3.634381] request_firmware_into_buf+0x4c/0x70 -------------------------------------------------------------------------- Fixes: 726e41097920 ("drivers: core: Remove glue dirs from sysfs earlier") Signed-off-by: Muchun Song <smuchun@gmail.com> Reviewed-by: Mukesh Ojha <mojha@codeaurora.org> Signed-off-by: Prateek Sood <prsood@codeaurora.org> Link: https://lore.kernel.org/r/20190727032122.24639-1-smuchun@gmail.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-07-27 03:21:22 +00:00
unsigned int ref;
/* see if we live in a "glue" directory */
if (!live_in_glue_dir(glue_dir, dev))
return;
sysfs: driver core: Fix glue dir race condition by gdp_mutex There is a race condition when removing glue directory. It can be reproduced in following test: path 1: Add first child device device_add() get_device_parent() /*find parent from glue_dirs.list*/ list_for_each_entry(k, &dev->class->p->glue_dirs.list, entry) if (k->parent == parent_kobj) { kobj = kobject_get(k); break; } .... class_dir_create_and_add() path2: Remove last child device under glue dir device_del() cleanup_device_parent() cleanup_glue_dir() kobject_put(glue_dir); If path2 has been called cleanup_glue_dir(), but not call kobject_put(glue_dir), the glue dir is still in parent's kset list. Meanwhile, path1 find the glue dir from the glue_dirs.list. Path2 may release glue dir before path1 call kobject_get(). So kernel will report the warning and bug_on. This is a "classic" problem we have of a kref in a list that can be found while the last instance could be removed at the same time. This patch reuse gdp_mutex to fix this race condition. The following calltrace is captured in kernel 3.4, but the latest kernel still has this bug. ----------------------------------------------------- <4>[ 3965.441471] WARNING: at ...include/linux/kref.h:41 kobject_get+0x33/0x40() <4>[ 3965.441474] Hardware name: Romley <4>[ 3965.441475] Modules linked in: isd_iop(O) isd_xda(O)... ... <4>[ 3965.441605] Call Trace: <4>[ 3965.441611] [<ffffffff8103717a>] warn_slowpath_common+0x7a/0xb0 <4>[ 3965.441615] [<ffffffff810371c5>] warn_slowpath_null+0x15/0x20 <4>[ 3965.441618] [<ffffffff81215963>] kobject_get+0x33/0x40 <4>[ 3965.441624] [<ffffffff812d1e45>] get_device_parent.isra.11+0x135/0x1f0 <4>[ 3965.441627] [<ffffffff812d22d4>] device_add+0xd4/0x6d0 <4>[ 3965.441631] [<ffffffff812d0dbc>] ? dev_set_name+0x3c/0x40 .... <2>[ 3965.441912] kernel BUG at ..../fs/sysfs/group.c:65! <4>[ 3965.441915] invalid opcode: 0000 [#1] SMP ... <4>[ 3965.686743] [<ffffffff811a677e>] sysfs_create_group+0xe/0x10 <4>[ 3965.686748] [<ffffffff810cfb04>] blk_trace_init_sysfs+0x14/0x20 <4>[ 3965.686753] [<ffffffff811fcabb>] blk_register_queue+0x3b/0x120 <4>[ 3965.686756] [<ffffffff812030bc>] add_disk+0x1cc/0x490 .... ------------------------------------------------------- Signed-off-by: Yijing Wang <wangyijing@huawei.com> Signed-off-by: Weng Meiling <wengmeiling.weng@huawei.com> Cc: <stable@vger.kernel.org> #3.4+ Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2014-11-07 04:05:49 +00:00
mutex_lock(&gdp_mutex);
driver core: Fix use-after-free and double free on glue directory There is a race condition between removing glue directory and adding a new device under the glue dir. It can be reproduced in following test: CPU1: CPU2: device_add() get_device_parent() class_dir_create_and_add() kobject_add_internal() create_dir() // create glue_dir device_add() get_device_parent() kobject_get() // get glue_dir device_del() cleanup_glue_dir() kobject_del(glue_dir) kobject_add() kobject_add_internal() create_dir() // in glue_dir sysfs_create_dir_ns() kernfs_create_dir_ns(sd) sysfs_remove_dir() // glue_dir->sd=NULL sysfs_put() // free glue_dir->sd // sd is freed kernfs_new_node(sd) kernfs_get(glue_dir) kernfs_add_one() kernfs_put() Before CPU1 remove last child device under glue dir, if CPU2 add a new device under glue dir, the glue_dir kobject reference count will be increase to 2 via kobject_get() in get_device_parent(). And CPU2 has been called kernfs_create_dir_ns(), but not call kernfs_new_node(). Meanwhile, CPU1 call sysfs_remove_dir() and sysfs_put(). This result in glue_dir->sd is freed and it's reference count will be 0. Then CPU2 call kernfs_get(glue_dir) will trigger a warning in kernfs_get() and increase it's reference count to 1. Because glue_dir->sd is freed by CPU1, the next call kernfs_add_one() by CPU2 will fail(This is also use-after-free) and call kernfs_put() to decrease reference count. Because the reference count is decremented to 0, it will also call kmem_cache_free() to free the glue_dir->sd again. This will result in double free. In order to avoid this happening, we also should make sure that kernfs_node for glue_dir is released in CPU1 only when refcount for glue_dir kobj is 1 to fix this race. The following calltrace is captured in kernel 4.14 with the following patch applied: commit 726e41097920 ("drivers: core: Remove glue dirs from sysfs earlier") -------------------------------------------------------------------------- [ 3.633703] WARNING: CPU: 4 PID: 513 at .../fs/kernfs/dir.c:494 Here is WARN_ON(!atomic_read(&kn->count) in kernfs_get(). .... [ 3.633986] Call trace: [ 3.633991] kernfs_create_dir_ns+0xa8/0xb0 [ 3.633994] sysfs_create_dir_ns+0x54/0xe8 [ 3.634001] kobject_add_internal+0x22c/0x3f0 [ 3.634005] kobject_add+0xe4/0x118 [ 3.634011] device_add+0x200/0x870 [ 3.634017] _request_firmware+0x958/0xc38 [ 3.634020] request_firmware_into_buf+0x4c/0x70 .... [ 3.634064] kernel BUG at .../mm/slub.c:294! Here is BUG_ON(object == fp) in set_freepointer(). .... [ 3.634346] Call trace: [ 3.634351] kmem_cache_free+0x504/0x6b8 [ 3.634355] kernfs_put+0x14c/0x1d8 [ 3.634359] kernfs_create_dir_ns+0x88/0xb0 [ 3.634362] sysfs_create_dir_ns+0x54/0xe8 [ 3.634366] kobject_add_internal+0x22c/0x3f0 [ 3.634370] kobject_add+0xe4/0x118 [ 3.634374] device_add+0x200/0x870 [ 3.634378] _request_firmware+0x958/0xc38 [ 3.634381] request_firmware_into_buf+0x4c/0x70 -------------------------------------------------------------------------- Fixes: 726e41097920 ("drivers: core: Remove glue dirs from sysfs earlier") Signed-off-by: Muchun Song <smuchun@gmail.com> Reviewed-by: Mukesh Ojha <mojha@codeaurora.org> Signed-off-by: Prateek Sood <prsood@codeaurora.org> Link: https://lore.kernel.org/r/20190727032122.24639-1-smuchun@gmail.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-07-27 03:21:22 +00:00
/**
* There is a race condition between removing glue directory
* and adding a new device under the glue directory.
*
* CPU1: CPU2:
*
* device_add()
* get_device_parent()
* class_dir_create_and_add()
* kobject_add_internal()
* create_dir() // create glue_dir
*
* device_add()
* get_device_parent()
* kobject_get() // get glue_dir
*
* device_del()
* cleanup_glue_dir()
* kobject_del(glue_dir)
*
* kobject_add()
* kobject_add_internal()
* create_dir() // in glue_dir
* sysfs_create_dir_ns()
* kernfs_create_dir_ns(sd)
*
* sysfs_remove_dir() // glue_dir->sd=NULL
* sysfs_put() // free glue_dir->sd
*
* // sd is freed
* kernfs_new_node(sd)
* kernfs_get(glue_dir)
* kernfs_add_one()
* kernfs_put()
*
* Before CPU1 remove last child device under glue dir, if CPU2 add
* a new device under glue dir, the glue_dir kobject reference count
* will be increase to 2 in kobject_get(k). And CPU2 has been called
* kernfs_create_dir_ns(). Meanwhile, CPU1 call sysfs_remove_dir()
* and sysfs_put(). This result in glue_dir->sd is freed.
*
* Then the CPU2 will see a stale "empty" but still potentially used
* glue dir around in kernfs_new_node().
*
* In order to avoid this happening, we also should make sure that
* kernfs_node for glue_dir is released in CPU1 only when refcount
* for glue_dir kobj is 1.
*/
ref = kref_read(&glue_dir->kref);
if (!kobject_has_children(glue_dir) && !--ref)
drivers: core: Remove glue dirs from sysfs earlier For devices with a class, we create a "glue" directory between the parent device and the new device with the class name. This directory is never "explicitely" removed when empty however, this is left to the implicit sysfs removal done by kobject_release() when the object loses its last reference via kobject_put(). This is problematic because as long as it's not been removed from sysfs, it is still present in the class kset and in sysfs directory structure. The presence in the class kset exposes a use after free bug fixed by the previous patch, but the presence in sysfs means that until the kobject is released, which can take a while (especially with kobject debugging), any attempt at re-creating such as binding a new device for that class/parent pair, will result in a sysfs duplicate file name error. This fixes it by instead doing an explicit kobject_del() when the glue dir is empty, by keeping track of the number of child devices of the gluedir. This is made easy by the fact that all glue dir operations are done with a global mutex, and there's already a function (cleanup_glue_dir) called in all the right places taking that mutex that can be enhanced for this. It appears that this was in fact the intent of the function, but the implementation was wrong. Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Acked-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2018-07-10 00:29:10 +00:00
kobject_del(glue_dir);
kobject_put(glue_dir);
sysfs: driver core: Fix glue dir race condition by gdp_mutex There is a race condition when removing glue directory. It can be reproduced in following test: path 1: Add first child device device_add() get_device_parent() /*find parent from glue_dirs.list*/ list_for_each_entry(k, &dev->class->p->glue_dirs.list, entry) if (k->parent == parent_kobj) { kobj = kobject_get(k); break; } .... class_dir_create_and_add() path2: Remove last child device under glue dir device_del() cleanup_device_parent() cleanup_glue_dir() kobject_put(glue_dir); If path2 has been called cleanup_glue_dir(), but not call kobject_put(glue_dir), the glue dir is still in parent's kset list. Meanwhile, path1 find the glue dir from the glue_dirs.list. Path2 may release glue dir before path1 call kobject_get(). So kernel will report the warning and bug_on. This is a "classic" problem we have of a kref in a list that can be found while the last instance could be removed at the same time. This patch reuse gdp_mutex to fix this race condition. The following calltrace is captured in kernel 3.4, but the latest kernel still has this bug. ----------------------------------------------------- <4>[ 3965.441471] WARNING: at ...include/linux/kref.h:41 kobject_get+0x33/0x40() <4>[ 3965.441474] Hardware name: Romley <4>[ 3965.441475] Modules linked in: isd_iop(O) isd_xda(O)... ... <4>[ 3965.441605] Call Trace: <4>[ 3965.441611] [<ffffffff8103717a>] warn_slowpath_common+0x7a/0xb0 <4>[ 3965.441615] [<ffffffff810371c5>] warn_slowpath_null+0x15/0x20 <4>[ 3965.441618] [<ffffffff81215963>] kobject_get+0x33/0x40 <4>[ 3965.441624] [<ffffffff812d1e45>] get_device_parent.isra.11+0x135/0x1f0 <4>[ 3965.441627] [<ffffffff812d22d4>] device_add+0xd4/0x6d0 <4>[ 3965.441631] [<ffffffff812d0dbc>] ? dev_set_name+0x3c/0x40 .... <2>[ 3965.441912] kernel BUG at ..../fs/sysfs/group.c:65! <4>[ 3965.441915] invalid opcode: 0000 [#1] SMP ... <4>[ 3965.686743] [<ffffffff811a677e>] sysfs_create_group+0xe/0x10 <4>[ 3965.686748] [<ffffffff810cfb04>] blk_trace_init_sysfs+0x14/0x20 <4>[ 3965.686753] [<ffffffff811fcabb>] blk_register_queue+0x3b/0x120 <4>[ 3965.686756] [<ffffffff812030bc>] add_disk+0x1cc/0x490 .... ------------------------------------------------------- Signed-off-by: Yijing Wang <wangyijing@huawei.com> Signed-off-by: Weng Meiling <wengmeiling.weng@huawei.com> Cc: <stable@vger.kernel.org> #3.4+ Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2014-11-07 04:05:49 +00:00
mutex_unlock(&gdp_mutex);
}
static int device_add_class_symlinks(struct device *dev)
{
struct device_node *of_node = dev_of_node(dev);
int error;
if (of_node) {
error = sysfs_create_link(&dev->kobj, of_node_kobj(of_node), "of_node");
if (error)
dev_warn(dev, "Error %d creating of_node link\n",error);
/* An error here doesn't warrant bringing down the device */
}
if (!dev->class)
return 0;
error = sysfs_create_link(&dev->kobj,
&dev->class->p->subsys.kobj,
"subsystem");
if (error)
goto out_devnode;
if (dev->parent && device_is_not_partition(dev)) {
error = sysfs_create_link(&dev->kobj, &dev->parent->kobj,
"device");
if (error)
goto out_subsys;
}
#ifdef CONFIG_BLOCK
/* /sys/block has directories and does not need symlinks */
if (sysfs_deprecated && dev->class == &block_class)
return 0;
#endif
/* link in the class directory pointing to the device */
error = sysfs_create_link(&dev->class->p->subsys.kobj,
&dev->kobj, dev_name(dev));
if (error)
goto out_device;
return 0;
out_device:
sysfs_remove_link(&dev->kobj, "device");
out_subsys:
sysfs_remove_link(&dev->kobj, "subsystem");
out_devnode:
sysfs_remove_link(&dev->kobj, "of_node");
return error;
}
static void device_remove_class_symlinks(struct device *dev)
{
if (dev_of_node(dev))
sysfs_remove_link(&dev->kobj, "of_node");
if (!dev->class)
return;
if (dev->parent && device_is_not_partition(dev))
sysfs_remove_link(&dev->kobj, "device");
sysfs_remove_link(&dev->kobj, "subsystem");
#ifdef CONFIG_BLOCK
if (sysfs_deprecated && dev->class == &block_class)
return;
#endif
sysfs_delete_link(&dev->class->p->subsys.kobj, &dev->kobj, dev_name(dev));
}
/**
* dev_set_name - set a device name
* @dev: device
* @fmt: format string for the device's name
*/
int dev_set_name(struct device *dev, const char *fmt, ...)
{
va_list vargs;
int err;
va_start(vargs, fmt);
err = kobject_set_name_vargs(&dev->kobj, fmt, vargs);
va_end(vargs);
return err;
}
EXPORT_SYMBOL_GPL(dev_set_name);
/**
* device_to_dev_kobj - select a /sys/dev/ directory for the device
* @dev: device
*
* By default we select char/ for new entries. Setting class->dev_obj
* to NULL prevents an entry from being created. class->dev_kobj must
* be set (or cleared) before any devices are registered to the class
* otherwise device_create_sys_dev_entry() and
* device_remove_sys_dev_entry() will disagree about the presence of
* the link.
*/
static struct kobject *device_to_dev_kobj(struct device *dev)
{
struct kobject *kobj;
if (dev->class)
kobj = dev->class->dev_kobj;
else
kobj = sysfs_dev_char_kobj;
return kobj;
}
static int device_create_sys_dev_entry(struct device *dev)
{
struct kobject *kobj = device_to_dev_kobj(dev);
int error = 0;
char devt_str[15];
if (kobj) {
format_dev_t(devt_str, dev->devt);
error = sysfs_create_link(kobj, &dev->kobj, devt_str);
}
return error;
}
static void device_remove_sys_dev_entry(struct device *dev)
{
struct kobject *kobj = device_to_dev_kobj(dev);
char devt_str[15];
if (kobj) {
format_dev_t(devt_str, dev->devt);
sysfs_remove_link(kobj, devt_str);
}
}
static int device_private_init(struct device *dev)
{
dev->p = kzalloc(sizeof(*dev->p), GFP_KERNEL);
if (!dev->p)
return -ENOMEM;
dev->p->device = dev;
klist_init(&dev->p->klist_children, klist_children_get,
klist_children_put);
INIT_LIST_HEAD(&dev->p->deferred_probe);
return 0;
}
/**
* device_add - add device to device hierarchy.
* @dev: device.
*
* This is part 2 of device_register(), though may be called
* separately _iff_ device_initialize() has been called separately.
*
* This adds @dev to the kobject hierarchy via kobject_add(), adds it
* to the global and sibling lists for the device, then
* adds it to the other relevant subsystems of the driver model.
*
* Do not call this routine or device_register() more than once for
* any device structure. The driver model core is not designed to work
* with devices that get unregistered and then spring back to life.
* (Among other things, it's very hard to guarantee that all references
* to the previous incarnation of @dev have been dropped.) Allocate
* and register a fresh new struct device instead.
*
* NOTE: _Never_ directly free @dev after calling this function, even
* if it returned an error! Always use put_device() to give up your
* reference instead.
*
* Rule of thumb is: if device_add() succeeds, you should call
* device_del() when you want to get rid of it. If device_add() has
* *not* succeeded, use *only* put_device() to drop the reference
* count.
*/
int device_add(struct device *dev)
{
struct device *parent;
struct kobject *kobj;
struct class_interface *class_intf;
int error = -EINVAL;
struct kobject *glue_dir = NULL;
dev = get_device(dev);
if (!dev)
goto done;
if (!dev->p) {
error = device_private_init(dev);
if (error)
goto done;
}
/*
* for statically allocated devices, which should all be converted
* some day, we need to initialize the name. We prevent reading back
* the name, and force the use of dev_name()
*/
if (dev->init_name) {
dev_set_name(dev, "%s", dev->init_name);
dev->init_name = NULL;
}
/* subsystems can specify simple device enumeration */
if (!dev_name(dev) && dev->bus && dev->bus->dev_name)
dev_set_name(dev, "%s%u", dev->bus->dev_name, dev->id);
if (!dev_name(dev)) {
error = -EINVAL;
goto name_error;
}
pr_debug("device: '%s': %s\n", dev_name(dev), __func__);
parent = get_device(dev->parent);
kobj = get_device_parent(dev, parent);
if (IS_ERR(kobj)) {
error = PTR_ERR(kobj);
goto parent_error;
}
if (kobj)
dev->kobj.parent = kobj;
/* use parent numa_node */
if (parent && (dev_to_node(dev) == NUMA_NO_NODE))
set_dev_node(dev, dev_to_node(parent));
/* first, register with generic layer. */
/* we require the name to be set before, and pass NULL */
error = kobject_add(&dev->kobj, dev->kobj.parent, NULL);
if (error) {
glue_dir = get_glue_dir(dev);
goto Error;
}
/* notify platform of device entry */
device_platform_notify(dev);
error = device_create_file(dev, &dev_attr_uevent);
if (error)
goto attrError;
error = device_add_class_symlinks(dev);
if (error)
goto SymlinkError;
error = device_add_attrs(dev);
if (error)
goto AttrsError;
error = bus_add_device(dev);
if (error)
goto BusError;
error = dpm_sysfs_add(dev);
if (error)
goto DPMError;
device_pm_add(dev);
if (MAJOR(dev->devt)) {
error = device_create_file(dev, &dev_attr_dev);
if (error)
goto DevAttrError;
error = device_create_sys_dev_entry(dev);
if (error)
goto SysEntryError;
devtmpfs_create_node(dev);
}
/* Notify clients of device addition. This call must come
* after dpm_sysfs_add() and before kobject_uevent().
*/
if (dev->bus)
blocking_notifier_call_chain(&dev->bus->p->bus_notifier,
BUS_NOTIFY_ADD_DEVICE, dev);
kobject_uevent(&dev->kobj, KOBJ_ADD);
/*
* Check if any of the other devices (consumers) have been waiting for
* this device (supplier) to be added so that they can create a device
* link to it.
*
* This needs to happen after device_pm_add() because device_link_add()
* requires the supplier be registered before it's called.
*
* But this also needs to happen before bus_probe_device() to make sure
* waiting consumers can link to it before the driver is bound to the
* device and the driver sync_state callback is called for this device.
*/
if (dev->fwnode && !dev->fwnode->dev) {
dev->fwnode->dev = dev;
fw_devlink_link_device(dev);
}
bus_probe_device(dev);
driver core: Improve fw_devlink & deferred_probe_timeout interaction deferred_probe_timeout kernel commandline parameter allows probing of consumer devices if the supplier devices don't have any drivers. fw_devlink=on will indefintely block probe() calls on a device if all its suppliers haven't probed successfully. This completely skips calls to driver_deferred_probe_check_state() since that's only called when a .probe() function calls framework APIs. So fw_devlink=on breaks deferred_probe_timeout. deferred_probe_timeout in its current state also ignores a lot of information that's now available to the kernel. It assumes all suppliers that haven't probed when the timer expires (or when initcalls are done on a static kernel) will never probe and fails any calls to acquire resources from these unprobed suppliers. However, this assumption by deferred_probe_timeout isn't true under many conditions. For example: - If the consumer happens to be before the supplier in the deferred probe list. - If the supplier itself is waiting on its supplier to probe. This patch fixes both these issues by relaxing device links between devices only if the supplier doesn't have any driver that could match with (NOT bound to) the supplier device. This way, we only fail attempts to acquire resources from suppliers that truly don't have any driver vs suppliers that just happen to not have probed yet. Signed-off-by: Saravana Kannan <saravanak@google.com> Link: https://lore.kernel.org/r/20210402040342.2944858-3-saravanak@google.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2021-04-02 04:03:41 +00:00
/*
* If all driver registration is done and a newly added device doesn't
* match with any driver, don't block its consumers from probing in
* case the consumer device is able to operate without this supplier.
*/
if (dev->fwnode && fw_devlink_drv_reg_done && !dev->can_match)
fw_devlink_unblock_consumers(dev);
if (parent)
klist_add_tail(&dev->p->knode_parent,
&parent->p->klist_children);
if (dev->class) {
mutex_lock(&dev->class->p->mutex);
/* tie the class to the device */
klist_add_tail(&dev->p->knode_class,
&dev->class->p->klist_devices);
/* notify any interfaces that the device is here */
list_for_each_entry(class_intf,
&dev->class->p->interfaces, node)
if (class_intf->add_dev)
class_intf->add_dev(dev, class_intf);
mutex_unlock(&dev->class->p->mutex);
}
done:
put_device(dev);
return error;
SysEntryError:
if (MAJOR(dev->devt))
device_remove_file(dev, &dev_attr_dev);
DevAttrError:
device_pm_remove(dev);
dpm_sysfs_remove(dev);
DPMError:
bus_remove_device(dev);
BusError:
device_remove_attrs(dev);
AttrsError:
device_remove_class_symlinks(dev);
SymlinkError:
device_remove_file(dev, &dev_attr_uevent);
attrError:
device_platform_notify_remove(dev);
kobject_uevent(&dev->kobj, KOBJ_REMOVE);
glue_dir = get_glue_dir(dev);
kobject_del(&dev->kobj);
Error:
cleanup_glue_dir(dev, glue_dir);
parent_error:
put_device(parent);
name_error:
kfree(dev->p);
dev->p = NULL;
goto done;
}
EXPORT_SYMBOL_GPL(device_add);
/**
* device_register - register a device with the system.
* @dev: pointer to the device structure
*
* This happens in two clean steps - initialize the device
* and add it to the system. The two steps can be called
* separately, but this is the easiest and most common.
* I.e. you should only call the two helpers separately if
* have a clearly defined need to use and refcount the device
* before it is added to the hierarchy.
*
* For more information, see the kerneldoc for device_initialize()
* and device_add().
*
* NOTE: _Never_ directly free @dev after calling this function, even
* if it returned an error! Always use put_device() to give up the
* reference initialized in this function instead.
*/
int device_register(struct device *dev)
{
device_initialize(dev);
return device_add(dev);
}
EXPORT_SYMBOL_GPL(device_register);
/**
* get_device - increment reference count for device.
* @dev: device.
*
* This simply forwards the call to kobject_get(), though
* we do take care to provide for the case that we get a NULL
* pointer passed in.
*/
struct device *get_device(struct device *dev)
{
return dev ? kobj_to_dev(kobject_get(&dev->kobj)) : NULL;
}
EXPORT_SYMBOL_GPL(get_device);
/**
* put_device - decrement reference count.
* @dev: device in question.
*/
void put_device(struct device *dev)
{
/* might_sleep(); */
if (dev)
kobject_put(&dev->kobj);
}
EXPORT_SYMBOL_GPL(put_device);
drivers/base: Introduce kill_device() The libnvdimm subsystem arranges for devices to be destroyed as a result of a sysfs operation. Since device_unregister() cannot be called from an actively running sysfs attribute of the same device libnvdimm arranges for device_unregister() to be performed in an out-of-line async context. The driver core maintains a 'dead' state for coordinating its own racing async registration / de-registration requests. Rather than add local 'dead' state tracking infrastructure to libnvdimm device objects, export the existing state tracking via a new kill_device() helper. The kill_device() helper simply marks the device as dead, i.e. that it is on its way to device_del(), or returns that the device was already dead. This can be used in advance of calling device_unregister() for subsystems like libnvdimm that might need to handle multiple user threads racing to delete a device. This refactoring does not change any behavior, but it is a pre-requisite for follow-on fixes and therefore marked for -stable. Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: "Rafael J. Wysocki" <rafael@kernel.org> Fixes: 4d88a97aa9e8 ("libnvdimm, nvdimm: dimm driver and base libnvdimm device-driver...") Cc: <stable@vger.kernel.org> Tested-by: Jane Chu <jane.chu@oracle.com> Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Link: https://lore.kernel.org/r/156341207332.292348.14959761496009347574.stgit@dwillia2-desk3.amr.corp.intel.com Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2019-07-18 01:07:53 +00:00
bool kill_device(struct device *dev)
{
/*
* Require the device lock and set the "dead" flag to guarantee that
* the update behavior is consistent with the other bitfields near
* it and that we cannot have an asynchronous probe routine trying
* to run while we are tearing out the bus/class/sysfs from
* underneath the device.
*/
device_lock_assert(dev);
drivers/base: Introduce kill_device() The libnvdimm subsystem arranges for devices to be destroyed as a result of a sysfs operation. Since device_unregister() cannot be called from an actively running sysfs attribute of the same device libnvdimm arranges for device_unregister() to be performed in an out-of-line async context. The driver core maintains a 'dead' state for coordinating its own racing async registration / de-registration requests. Rather than add local 'dead' state tracking infrastructure to libnvdimm device objects, export the existing state tracking via a new kill_device() helper. The kill_device() helper simply marks the device as dead, i.e. that it is on its way to device_del(), or returns that the device was already dead. This can be used in advance of calling device_unregister() for subsystems like libnvdimm that might need to handle multiple user threads racing to delete a device. This refactoring does not change any behavior, but it is a pre-requisite for follow-on fixes and therefore marked for -stable. Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: "Rafael J. Wysocki" <rafael@kernel.org> Fixes: 4d88a97aa9e8 ("libnvdimm, nvdimm: dimm driver and base libnvdimm device-driver...") Cc: <stable@vger.kernel.org> Tested-by: Jane Chu <jane.chu@oracle.com> Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Link: https://lore.kernel.org/r/156341207332.292348.14959761496009347574.stgit@dwillia2-desk3.amr.corp.intel.com Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2019-07-18 01:07:53 +00:00
if (dev->p->dead)
return false;
dev->p->dead = true;
return true;
}
EXPORT_SYMBOL_GPL(kill_device);
/**
* device_del - delete device from system.
* @dev: device.
*
* This is the first part of the device unregistration
* sequence. This removes the device from the lists we control
* from here, has it removed from the other driver model
* subsystems it was added to in device_add(), and removes it
* from the kobject hierarchy.
*
* NOTE: this should be called manually _iff_ device_add() was
* also called manually.
*/
void device_del(struct device *dev)
{
struct device *parent = dev->parent;
struct kobject *glue_dir = NULL;
struct class_interface *class_intf;
unsigned int noio_flag;
device_lock(dev);
drivers/base: Introduce kill_device() The libnvdimm subsystem arranges for devices to be destroyed as a result of a sysfs operation. Since device_unregister() cannot be called from an actively running sysfs attribute of the same device libnvdimm arranges for device_unregister() to be performed in an out-of-line async context. The driver core maintains a 'dead' state for coordinating its own racing async registration / de-registration requests. Rather than add local 'dead' state tracking infrastructure to libnvdimm device objects, export the existing state tracking via a new kill_device() helper. The kill_device() helper simply marks the device as dead, i.e. that it is on its way to device_del(), or returns that the device was already dead. This can be used in advance of calling device_unregister() for subsystems like libnvdimm that might need to handle multiple user threads racing to delete a device. This refactoring does not change any behavior, but it is a pre-requisite for follow-on fixes and therefore marked for -stable. Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: "Rafael J. Wysocki" <rafael@kernel.org> Fixes: 4d88a97aa9e8 ("libnvdimm, nvdimm: dimm driver and base libnvdimm device-driver...") Cc: <stable@vger.kernel.org> Tested-by: Jane Chu <jane.chu@oracle.com> Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Link: https://lore.kernel.org/r/156341207332.292348.14959761496009347574.stgit@dwillia2-desk3.amr.corp.intel.com Signed-off-by: Dan Williams <dan.j.williams@intel.com>
2019-07-18 01:07:53 +00:00
kill_device(dev);
device_unlock(dev);
if (dev->fwnode && dev->fwnode->dev == dev)
dev->fwnode->dev = NULL;
/* Notify clients of device removal. This call must come
* before dpm_sysfs_remove().
*/
noio_flag = memalloc_noio_save();
if (dev->bus)
blocking_notifier_call_chain(&dev->bus->p->bus_notifier,
BUS_NOTIFY_DEL_DEVICE, dev);
driver core: Functional dependencies tracking support Currently, there is a problem with taking functional dependencies between devices into account. What I mean by a "functional dependency" is when the driver of device B needs device A to be functional and (generally) its driver to be present in order to work properly. This has certain consequences for power management (suspend/resume and runtime PM ordering) and shutdown ordering of these devices. In general, it also implies that the driver of A needs to be working for B to be probed successfully and it cannot be unbound from the device before the B's driver. Support for representing those functional dependencies between devices is added here to allow the driver core to track them and act on them in certain cases where applicable. The argument for doing that in the driver core is that there are quite a few distinct use cases involving device dependencies, they are relatively hard to get right in a driver (if one wants to address all of them properly) and it only gets worse if multiplied by the number of drivers potentially needing to do it. Morever, at least one case (asynchronous system suspend/resume) cannot be handled in a single driver at all, because it requires the driver of A to wait for B to suspend (during system suspend) and the driver of B to wait for A to resume (during system resume). For this reason, represent dependencies between devices as "links", with the help of struct device_link objects each containing pointers to the "linked" devices, a list node for each of them, status information, flags, and an RCU head for synchronization. Also add two new list heads, representing the lists of links to the devices that depend on the given one (consumers) and to the devices depended on by it (suppliers), and a "driver presence status" field (needed for figuring out initial states of device links) to struct device. The entire data structure consisting of all of the lists of link objects for all devices is protected by a mutex (for link object addition/removal and for list walks during device driver probing and removal) and by SRCU (for list walking in other case that will be introduced by subsequent change sets). If CONFIG_SRCU is not selected, however, an rwsem is used for protecting the entire data structure. In addition, each link object has an internal status field whose value reflects whether or not drivers are bound to the devices pointed to by the link or probing/removal of their drivers is in progress etc. That field is only modified under the device links mutex, but it may be read outside of it in some cases (introduced by subsequent change sets), so modifications of it are annotated with WRITE_ONCE(). New links are added by calling device_link_add() which takes three arguments: pointers to the devices in question and flags. In particular, if DL_FLAG_STATELESS is set in the flags, the link status is not to be taken into account for this link and the driver core will not manage it. In turn, if DL_FLAG_AUTOREMOVE is set in the flags, the driver core will remove the link automatically when the consumer device driver unbinds from it. One of the actions carried out by device_link_add() is to reorder the lists used for device shutdown and system suspend/resume to put the consumer device along with all of its children and all of its consumers (and so on, recursively) to the ends of those lists in order to ensure the right ordering between all of the supplier and consumer devices. For this reason, it is not possible to create a link between two devices if the would-be supplier device already depends on the would-be consumer device as either a direct descendant of it or a consumer of one of its direct descendants or one of its consumers and so on. There are two types of link objects, persistent and non-persistent. The persistent ones stay around until one of the target devices is deleted, while the non-persistent ones are removed automatically when the consumer driver unbinds from its device (ie. they are assumed to be valid only as long as the consumer device has a driver bound to it). Persistent links are created by default and non-persistent links are created when the DL_FLAG_AUTOREMOVE flag is passed to device_link_add(). Both persistent and non-persistent device links can be deleted with an explicit call to device_link_del(). Links created without the DL_FLAG_STATELESS flag set are managed by the driver core using a simple state machine. There are 5 states each link can be in: DORMANT (unused), AVAILABLE (the supplier driver is present and functional), CONSUMER_PROBE (the consumer driver is probing), ACTIVE (both supplier and consumer drivers are present and functional), and SUPPLIER_UNBIND (the supplier driver is unbinding). The driver core updates the link state automatically depending on what happens to the linked devices and for each link state specific actions are taken in addition to that. For example, if the supplier driver unbinds from its device, the driver core will also unbind the drivers of all of its consumers automatically under the assumption that they cannot function properly without the supplier. Analogously, the driver core will only allow the consumer driver to bind to its device if the supplier driver is present and functional (ie. the link is in the AVAILABLE state). If that's not the case, it will rely on the existing deferred probing mechanism to wait for the supplier driver to become available. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2016-10-30 16:32:16 +00:00
dpm_sysfs_remove(dev);
if (parent)
klist_del(&dev->p->knode_parent);
if (MAJOR(dev->devt)) {
Driver Core: devtmpfs - kernel-maintained tmpfs-based /dev Devtmpfs lets the kernel create a tmpfs instance called devtmpfs very early at kernel initialization, before any driver-core device is registered. Every device with a major/minor will provide a device node in devtmpfs. Devtmpfs can be changed and altered by userspace at any time, and in any way needed - just like today's udev-mounted tmpfs. Unmodified udev versions will run just fine on top of it, and will recognize an already existing kernel-created device node and use it. The default node permissions are root:root 0600. Proper permissions and user/group ownership, meaningful symlinks, all other policy still needs to be applied by userspace. If a node is created by devtmps, devtmpfs will remove the device node when the device goes away. If the device node was created by userspace, or the devtmpfs created node was replaced by userspace, it will no longer be removed by devtmpfs. If it is requested to auto-mount it, it makes init=/bin/sh work without any further userspace support. /dev will be fully populated and dynamic, and always reflect the current device state of the kernel. With the commonly used dynamic device numbers, it solves the problem where static devices nodes may point to the wrong devices. It is intended to make the initial bootup logic simpler and more robust, by de-coupling the creation of the inital environment, to reliably run userspace processes, from a complex userspace bootstrap logic to provide a working /dev. Signed-off-by: Kay Sievers <kay.sievers@vrfy.org> Signed-off-by: Jan Blunck <jblunck@suse.de> Tested-By: Harald Hoyer <harald@redhat.com> Tested-By: Scott James Remnant <scott@ubuntu.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2009-04-30 13:23:42 +00:00
devtmpfs_delete_node(dev);
device_remove_sys_dev_entry(dev);
device_remove_file(dev, &dev_attr_dev);
}
if (dev->class) {
device_remove_class_symlinks(dev);
mutex_lock(&dev->class->p->mutex);
/* notify any interfaces that the device is now gone */
list_for_each_entry(class_intf,
&dev->class->p->interfaces, node)
if (class_intf->remove_dev)
class_intf->remove_dev(dev, class_intf);
/* remove the device from the class list */
klist_del(&dev->p->knode_class);
mutex_unlock(&dev->class->p->mutex);
}
device_remove_file(dev, &dev_attr_uevent);
device_remove_attrs(dev);
bus_remove_device(dev);
driver core / PM: move the calling to device_pm_remove behind the calling to bus_remove_device We hit an hang issue when removing a mmc device on Medfield Android phone by sysfs interface. device_pm_remove will call pm_runtime_remove which would disable runtime PM of the device. After that pm_runtime_get* or pm_runtime_put* will be ignored. So if we disable the runtime PM before device really be removed, drivers' _remove callback may access HW even pm_runtime_get* fails. That is bad. Consider below call sequence when removing a device: device_del => device_pm_remove => class_intf->remove_dev(dev, class_intf) => pm_runtime_get_sync/put_sync => bus_remove_device => device_release_driver => pm_runtime_get_sync/put_sync remove_dev might call pm_runtime_get_sync/put_sync. Then, generic device_release_driver also calls pm_runtime_get_sync/put_sync. Since device_del => device_pm_remove firstly, later _get_sync wouldn't really wake up the device. I git log -p to find the patch which moves the calling to device_pm_remove ahead. It's below patch: commit 775b64d2b6ca37697de925f70799c710aab5849a Author: Rafael J. Wysocki <rjw@sisk.pl> Date: Sat Jan 12 20:40:46 2008 +0100 PM: Acquire device locks on suspend This patch reorganizes the way suspend and resume notifications are sent to drivers. The major changes are that now the PM core acquires every device semaphore before calling the methods, and calls to device_add() during suspends will fail, while calls to device_del() during suspends will block. It also provides a way to safely remove a suspended device with the help of the PM core, by using the device_pm_schedule_removal() callback introduced specifically for this purpose, and updates two drivers (msr and cpuid) that need to use it. As device_pm_schedule_removal is deleted by another patch, we need also revert other parts of the patch, i.e. move the calling of device_pm_remove after the calling to bus_remove_device. Signed-off-by: LongX Zhang <longx.zhang@intel.com> Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
2012-10-24 22:21:28 +00:00
device_pm_remove(dev);
drivercore: Add driver probe deferral mechanism Allow drivers to report at probe time that they cannot get all the resources required by the device, and should be retried at a later time. This should completely solve the problem of getting devices initialized in the right order. Right now this is mostly handled by mucking about with initcall ordering which is a complete hack, and doesn't even remotely handle the case where device drivers are in modules. This approach completely sidesteps the issues by allowing driver registration to occur in any order, and any driver can request to be retried after a few more other drivers get probed. v4: - Integrate Manjunath's addition of a separate workqueue - Change -EAGAIN to -EPROBE_DEFER for drivers to trigger deferral - Update comment blocks to reflect how the code really works v3: - Hold off workqueue scheduling until late_initcall so that the bulk of driver probes are complete before we start retrying deferred devices. - Tested with simple use cases. Still needs more testing though. Using it to get rid of the gpio early_initcall madness, or to replace the ASoC internal probe deferral code would be ideal. v2: - added locking so it should no longer be utterly broken in that regard - remove device from deferred list at device_del time. - Still completely untested with any real use case, but has been boot tested. Signed-off-by: Grant Likely <grant.likely@secretlab.ca> Cc: Mark Brown <broonie@opensource.wolfsonmicro.com> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Dilan Lee <dilee@nvidia.com> Cc: Manjunath GKondaiah <manjunath.gkondaiah@linaro.org> Cc: Alan Stern <stern@rowland.harvard.edu> Cc: Tony Lindgren <tony@atomide.com> Cc: Alan Cox <alan@lxorguk.ukuu.org.uk> Reviewed-by: Mark Brown <broonie@opensource.wolfsonmicro.com> Acked-by: David Daney <david.daney@cavium.com> Reviewed-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-03-05 15:47:41 +00:00
driver_deferred_probe_del(dev);
device_platform_notify_remove(dev);
device_links_purge(dev);
if (dev->bus)
blocking_notifier_call_chain(&dev->bus->p->bus_notifier,
BUS_NOTIFY_REMOVED_DEVICE, dev);
kobject_uevent(&dev->kobj, KOBJ_REMOVE);
glue_dir = get_glue_dir(dev);
kobject_del(&dev->kobj);
cleanup_glue_dir(dev, glue_dir);
memalloc_noio_restore(noio_flag);
put_device(parent);
}
EXPORT_SYMBOL_GPL(device_del);
/**
* device_unregister - unregister device from system.
* @dev: device going away.
*
* We do this in two parts, like we do device_register(). First,
* we remove it from all the subsystems with device_del(), then
* we decrement the reference count via put_device(). If that
* is the final reference count, the device will be cleaned up
* via device_release() above. Otherwise, the structure will
* stick around until the final reference to the device is dropped.
*/
void device_unregister(struct device *dev)
{
pr_debug("device: '%s': %s\n", dev_name(dev), __func__);
device_del(dev);
put_device(dev);
}
EXPORT_SYMBOL_GPL(device_unregister);
static struct device *prev_device(struct klist_iter *i)
{
struct klist_node *n = klist_prev(i);
struct device *dev = NULL;
struct device_private *p;
if (n) {
p = to_device_private_parent(n);
dev = p->device;
}
return dev;
}
static struct device *next_device(struct klist_iter *i)
{
struct klist_node *n = klist_next(i);
struct device *dev = NULL;
struct device_private *p;
if (n) {
p = to_device_private_parent(n);
dev = p->device;
}
return dev;
}
/**
* device_get_devnode - path of device node file
* @dev: device
* @mode: returned file access mode
* @uid: returned file owner
* @gid: returned file group
* @tmp: possibly allocated string
*
* Return the relative path of a possible device node.
* Non-default names may need to allocate a memory to compose
* a name. This memory is returned in tmp and needs to be
* freed by the caller.
*/
const char *device_get_devnode(struct device *dev,
umode_t *mode, kuid_t *uid, kgid_t *gid,
const char **tmp)
{
char *s;
*tmp = NULL;
/* the device type may provide a specific name */
if (dev->type && dev->type->devnode)
*tmp = dev->type->devnode(dev, mode, uid, gid);
if (*tmp)
return *tmp;
/* the class may provide a specific name */
if (dev->class && dev->class->devnode)
*tmp = dev->class->devnode(dev, mode);
if (*tmp)
return *tmp;
/* return name without allocation, tmp == NULL */
if (strchr(dev_name(dev), '!') == NULL)
return dev_name(dev);
/* replace '!' in the name with '/' */
s = kstrdup(dev_name(dev), GFP_KERNEL);
if (!s)
return NULL;
strreplace(s, '!', '/');
return *tmp = s;
}
/**
* device_for_each_child - device child iterator.
* @parent: parent struct device.
* @fn: function to be called for each device.
* @data: data for the callback.
*
* Iterate over @parent's child devices, and call @fn for each,
* passing it @data.
*
* We check the return of @fn each time. If it returns anything
* other than 0, we break out and return that value.
*/
int device_for_each_child(struct device *parent, void *data,
int (*fn)(struct device *dev, void *data))
{
struct klist_iter i;
struct device *child;
int error = 0;
if (!parent->p)
return 0;
klist_iter_init(&parent->p->klist_children, &i);
while (!error && (child = next_device(&i)))
error = fn(child, data);
klist_iter_exit(&i);
return error;
}
EXPORT_SYMBOL_GPL(device_for_each_child);
/**
* device_for_each_child_reverse - device child iterator in reversed order.
* @parent: parent struct device.
* @fn: function to be called for each device.
* @data: data for the callback.
*
* Iterate over @parent's child devices, and call @fn for each,
* passing it @data.
*
* We check the return of @fn each time. If it returns anything
* other than 0, we break out and return that value.
*/
int device_for_each_child_reverse(struct device *parent, void *data,
int (*fn)(struct device *dev, void *data))
{
struct klist_iter i;
struct device *child;
int error = 0;
if (!parent->p)
return 0;
klist_iter_init(&parent->p->klist_children, &i);
while ((child = prev_device(&i)) && !error)
error = fn(child, data);
klist_iter_exit(&i);
return error;
}
EXPORT_SYMBOL_GPL(device_for_each_child_reverse);
/**
* device_find_child - device iterator for locating a particular device.
* @parent: parent struct device
* @match: Callback function to check device
* @data: Data to pass to match function
*
* This is similar to the device_for_each_child() function above, but it
* returns a reference to a device that is 'found' for later use, as
* determined by the @match callback.
*
* The callback should return 0 if the device doesn't match and non-zero
* if it does. If the callback returns non-zero and a reference to the
* current device can be obtained, this function will return to the caller
* and not iterate over any more devices.
*
* NOTE: you will need to drop the reference with put_device() after use.
*/
struct device *device_find_child(struct device *parent, void *data,
int (*match)(struct device *dev, void *data))
{
struct klist_iter i;
struct device *child;
if (!parent)
return NULL;
klist_iter_init(&parent->p->klist_children, &i);
while ((child = next_device(&i)))
if (match(child, data) && get_device(child))
break;
klist_iter_exit(&i);
return child;
}
EXPORT_SYMBOL_GPL(device_find_child);
/**
* device_find_child_by_name - device iterator for locating a child device.
* @parent: parent struct device
* @name: name of the child device
*
* This is similar to the device_find_child() function above, but it
* returns a reference to a device that has the name @name.
*
* NOTE: you will need to drop the reference with put_device() after use.
*/
struct device *device_find_child_by_name(struct device *parent,
const char *name)
{
struct klist_iter i;
struct device *child;
if (!parent)
return NULL;
klist_iter_init(&parent->p->klist_children, &i);
while ((child = next_device(&i)))
drivers/base: make device_find_child_by_name() compatible with sysfs inputs Use sysfs_streq() in device_find_child_by_name() to allow it to use a sysfs input string that might contain a trailing newline. The other "device by name" interfaces, {bus,driver,class}_find_device_by_name(), already account for sysfs strings. Signed-off-by: Dan Williams <dan.j.williams@intel.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Andy Lutomirski <luto@kernel.org> Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org> Cc: Ard Biesheuvel <ardb@kernel.org> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Ben Skeggs <bskeggs@redhat.com> Cc: Bjorn Helgaas <bhelgaas@google.com> Cc: Borislav Petkov <bp@alien8.de> Cc: Boris Ostrovsky <boris.ostrovsky@oracle.com> Cc: Brice Goglin <Brice.Goglin@inria.fr> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Daniel Vetter <daniel@ffwll.ch> Cc: Dave Hansen <dave.hansen@linux.intel.com> Cc: Dave Jiang <dave.jiang@intel.com> Cc: David Airlie <airlied@linux.ie> Cc: David Hildenbrand <david@redhat.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Hulk Robot <hulkci@huawei.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Ira Weiny <ira.weiny@intel.com> Cc: Jason Gunthorpe <jgg@mellanox.com> Cc: Jason Yan <yanaijie@huawei.com> Cc: Jeff Moyer <jmoyer@redhat.com> Cc: "Jérôme Glisse" <jglisse@redhat.com> Cc: Jia He <justin.he@arm.com> Cc: Joao Martins <joao.m.martins@oracle.com> Cc: Jonathan Cameron <Jonathan.Cameron@huawei.com> Cc: Juergen Gross <jgross@suse.com> Cc: kernel test robot <lkp@intel.com> Cc: Michael Ellerman <mpe@ellerman.id.au> Cc: Mike Rapoport <rppt@linux.ibm.com> Cc: Paul Mackerras <paulus@ozlabs.org> Cc: Pavel Tatashin <pasha.tatashin@soleen.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: "Rafael J. Wysocki" <rafael.j.wysocki@intel.com> Cc: Randy Dunlap <rdunlap@infradead.org> Cc: Stefano Stabellini <sstabellini@kernel.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Tom Lendacky <thomas.lendacky@amd.com> Cc: Vishal Verma <vishal.l.verma@intel.com> Cc: Vivek Goyal <vgoyal@redhat.com> Cc: Wei Yang <richard.weiyang@linux.alibaba.com> Cc: Will Deacon <will@kernel.org> Link: https://lkml.kernel.org/r/159643102106.4062302.12229802117645312104.stgit@dwillia2-desk3.amr.corp.intel.com Link: https://lkml.kernel.org/r/160106114576.30709.2960091665444712180.stgit@dwillia2-desk3.amr.corp.intel.com Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-10-13 23:50:18 +00:00
if (sysfs_streq(dev_name(child), name) && get_device(child))
break;
klist_iter_exit(&i);
return child;
}
EXPORT_SYMBOL_GPL(device_find_child_by_name);
static int match_any(struct device *dev, void *unused)
{
return 1;
}
/**
* device_find_any_child - device iterator for locating a child device, if any.
* @parent: parent struct device
*
* This is similar to the device_find_child() function above, but it
* returns a reference to a child device, if any.
*
* NOTE: you will need to drop the reference with put_device() after use.
*/
struct device *device_find_any_child(struct device *parent)
{
return device_find_child(parent, NULL, match_any);
}
EXPORT_SYMBOL_GPL(device_find_any_child);
int __init devices_init(void)
{
devices_kset = kset_create_and_add("devices", &device_uevent_ops, NULL);
if (!devices_kset)
return -ENOMEM;
dev_kobj = kobject_create_and_add("dev", NULL);
if (!dev_kobj)
goto dev_kobj_err;
sysfs_dev_block_kobj = kobject_create_and_add("block", dev_kobj);
if (!sysfs_dev_block_kobj)
goto block_kobj_err;
sysfs_dev_char_kobj = kobject_create_and_add("char", dev_kobj);
if (!sysfs_dev_char_kobj)
goto char_kobj_err;
return 0;
char_kobj_err:
kobject_put(sysfs_dev_block_kobj);
block_kobj_err:
kobject_put(dev_kobj);
dev_kobj_err:
kset_unregister(devices_kset);
return -ENOMEM;
}
Driver core: Add offline/online device operations In some cases, graceful hot-removal of devices is not possible, although in principle the devices in question support hotplug. For example, that may happen for the last CPU in the system or for memory modules holding kernel memory. In those cases it is nice to be able to check if the given device can be gracefully hot-removed before triggering a removal procedure that cannot be aborted or reversed. Unfortunately, however, the kernel currently doesn't provide any support for that. To address that deficiency, introduce support for offline and online operations that can be performed on devices, respectively, before a hot-removal and in case when it is necessary (or convenient) to put a device back online after a successful offline (that has not been followed by removal). The idea is that the offline will fail whenever the given device cannot be gracefully removed from the system and it will not be allowed to use the device after a successful offline (until a subsequent online) in analogy with the existing CPU offline/online mechanism. For now, the offline and online operations are introduced at the bus type level, as that should be sufficient for the most urgent use cases (CPUs and memory modules). In the future, however, the approach may be extended to cover some more complicated device offline/online scenarios involving device drivers etc. The lock_device_hotplug() and unlock_device_hotplug() functions are introduced because subsequent patches need to put larger pieces of code under device_hotplug_lock to prevent race conditions between device offline and removal from happening. Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Reviewed-by: Toshi Kani <toshi.kani@hp.com>
2013-05-02 20:15:29 +00:00
static int device_check_offline(struct device *dev, void *not_used)
{
int ret;
ret = device_for_each_child(dev, NULL, device_check_offline);
if (ret)
return ret;
return device_supports_offline(dev) && !dev->offline ? -EBUSY : 0;
}
/**
* device_offline - Prepare the device for hot-removal.
* @dev: Device to be put offline.
*
* Execute the device bus type's .offline() callback, if present, to prepare
* the device for a subsequent hot-removal. If that succeeds, the device must
* not be used until either it is removed or its bus type's .online() callback
* is executed.
*
* Call under device_hotplug_lock.
*/
int device_offline(struct device *dev)
{
int ret;
if (dev->offline_disabled)
return -EPERM;
ret = device_for_each_child(dev, NULL, device_check_offline);
if (ret)
return ret;
device_lock(dev);
if (device_supports_offline(dev)) {
if (dev->offline) {
ret = 1;
} else {
ret = dev->bus->offline(dev);
if (!ret) {
kobject_uevent(&dev->kobj, KOBJ_OFFLINE);
dev->offline = true;
}
}
}
device_unlock(dev);
return ret;
}
/**
* device_online - Put the device back online after successful device_offline().
* @dev: Device to be put back online.
*
* If device_offline() has been successfully executed for @dev, but the device
* has not been removed subsequently, execute its bus type's .online() callback
* to indicate that the device can be used again.
*
* Call under device_hotplug_lock.
*/
int device_online(struct device *dev)
{
int ret = 0;
device_lock(dev);
if (device_supports_offline(dev)) {
if (dev->offline) {
ret = dev->bus->online(dev);
if (!ret) {
kobject_uevent(&dev->kobj, KOBJ_ONLINE);
dev->offline = false;
}
} else {
ret = 1;
}
}
device_unlock(dev);
return ret;
}
struct root_device {
struct device dev;
struct module *owner;
};
static inline struct root_device *to_root_device(struct device *d)
{
return container_of(d, struct root_device, dev);
}
static void root_device_release(struct device *dev)
{
kfree(to_root_device(dev));
}
/**
* __root_device_register - allocate and register a root device
* @name: root device name
* @owner: owner module of the root device, usually THIS_MODULE
*
* This function allocates a root device and registers it
* using device_register(). In order to free the returned
* device, use root_device_unregister().
*
* Root devices are dummy devices which allow other devices
* to be grouped under /sys/devices. Use this function to
* allocate a root device and then use it as the parent of
* any device which should appear under /sys/devices/{name}
*
* The /sys/devices/{name} directory will also contain a
* 'module' symlink which points to the @owner directory
* in sysfs.
*
* Returns &struct device pointer on success, or ERR_PTR() on error.
*
* Note: You probably want to use root_device_register().
*/
struct device *__root_device_register(const char *name, struct module *owner)
{
struct root_device *root;
int err = -ENOMEM;
root = kzalloc(sizeof(struct root_device), GFP_KERNEL);
if (!root)
return ERR_PTR(err);
err = dev_set_name(&root->dev, "%s", name);
if (err) {
kfree(root);
return ERR_PTR(err);
}
root->dev.release = root_device_release;
err = device_register(&root->dev);
if (err) {
put_device(&root->dev);
return ERR_PTR(err);
}
#ifdef CONFIG_MODULES /* gotta find a "cleaner" way to do this */
if (owner) {
struct module_kobject *mk = &owner->mkobj;
err = sysfs_create_link(&root->dev.kobj, &mk->kobj, "module");
if (err) {
device_unregister(&root->dev);
return ERR_PTR(err);
}
root->owner = owner;
}
#endif
return &root->dev;
}
EXPORT_SYMBOL_GPL(__root_device_register);
/**
* root_device_unregister - unregister and free a root device
* @dev: device going away
*
* This function unregisters and cleans up a device that was created by
* root_device_register().
*/
void root_device_unregister(struct device *dev)
{
struct root_device *root = to_root_device(dev);
if (root->owner)
sysfs_remove_link(&root->dev.kobj, "module");
device_unregister(dev);
}
EXPORT_SYMBOL_GPL(root_device_unregister);
static void device_create_release(struct device *dev)
{
pr_debug("device: '%s': %s\n", dev_name(dev), __func__);
kfree(dev);
}
static __printf(6, 0) struct device *
device_create_groups_vargs(struct class *class, struct device *parent,
dev_t devt, void *drvdata,
const struct attribute_group **groups,
const char *fmt, va_list args)
{
struct device *dev = NULL;
int retval = -ENODEV;
if (class == NULL || IS_ERR(class))
goto error;
dev = kzalloc(sizeof(*dev), GFP_KERNEL);
if (!dev) {
retval = -ENOMEM;
goto error;
}
device_initialize(dev);
dev->devt = devt;
dev->class = class;
dev->parent = parent;
dev->groups = groups;
dev->release = device_create_release;
dev_set_drvdata(dev, drvdata);
retval = kobject_set_name_vargs(&dev->kobj, fmt, args);
if (retval)
goto error;
retval = device_add(dev);
if (retval)
goto error;
return dev;
error:
put_device(dev);
return ERR_PTR(retval);
}
/**
* device_create - creates a device and registers it with sysfs
* @class: pointer to the struct class that this device should be registered to
* @parent: pointer to the parent struct device of this new device, if any
* @devt: the dev_t for the char device to be added
* @drvdata: the data to be added to the device for callbacks
* @fmt: string for the device's name
*
* This function can be used by char device classes. A struct device
* will be created in sysfs, registered to the specified class.
*
* A "dev" file will be created, showing the dev_t for the device, if
* the dev_t is not 0,0.
* If a pointer to a parent struct device is passed in, the newly created
* struct device will be a child of that device in sysfs.
* The pointer to the struct device will be returned from the call.
* Any further sysfs files that might be required can be created using this
* pointer.
*
* Returns &struct device pointer on success, or ERR_PTR() on error.
*
* Note: the struct class passed to this function must have previously
* been created with a call to class_create().
*/
struct device *device_create(struct class *class, struct device *parent,
dev_t devt, void *drvdata, const char *fmt, ...)
{
va_list vargs;
struct device *dev;
va_start(vargs, fmt);
dev = device_create_groups_vargs(class, parent, devt, drvdata, NULL,
fmt, vargs);
va_end(vargs);
return dev;
}
EXPORT_SYMBOL_GPL(device_create);
/**
* device_create_with_groups - creates a device and registers it with sysfs
* @class: pointer to the struct class that this device should be registered to
* @parent: pointer to the parent struct device of this new device, if any
* @devt: the dev_t for the char device to be added
* @drvdata: the data to be added to the device for callbacks
* @groups: NULL-terminated list of attribute groups to be created
* @fmt: string for the device's name
*
* This function can be used by char device classes. A struct device
* will be created in sysfs, registered to the specified class.
* Additional attributes specified in the groups parameter will also
* be created automatically.
*
* A "dev" file will be created, showing the dev_t for the device, if
* the dev_t is not 0,0.
* If a pointer to a parent struct device is passed in, the newly created
* struct device will be a child of that device in sysfs.
* The pointer to the struct device will be returned from the call.
* Any further sysfs files that might be required can be created using this
* pointer.
*
* Returns &struct device pointer on success, or ERR_PTR() on error.
*
* Note: the struct class passed to this function must have previously
* been created with a call to class_create().
*/
struct device *device_create_with_groups(struct class *class,
struct device *parent, dev_t devt,
void *drvdata,
const struct attribute_group **groups,
const char *fmt, ...)
{
va_list vargs;
struct device *dev;
va_start(vargs, fmt);
dev = device_create_groups_vargs(class, parent, devt, drvdata, groups,
fmt, vargs);
va_end(vargs);
return dev;
}
EXPORT_SYMBOL_GPL(device_create_with_groups);
/**
* device_destroy - removes a device that was created with device_create()
* @class: pointer to the struct class that this device was registered with
* @devt: the dev_t of the device that was previously registered
*
* This call unregisters and cleans up a device that was created with a
* call to device_create().
*/
void device_destroy(struct class *class, dev_t devt)
{
struct device *dev;
dev = class_find_device_by_devt(class, devt);
if (dev) {
put_device(dev);
device_unregister(dev);
}
}
EXPORT_SYMBOL_GPL(device_destroy);
/**
* device_rename - renames a device
* @dev: the pointer to the struct device to be renamed
* @new_name: the new name of the device
*
* It is the responsibility of the caller to provide mutual
* exclusion between two different calls of device_rename
* on the same device to ensure that new_name is valid and
* won't conflict with other devices.
*
* Note: Don't call this function. Currently, the networking layer calls this
* function, but that will change. The following text from Kay Sievers offers
* some insight:
*
* Renaming devices is racy at many levels, symlinks and other stuff are not
* replaced atomically, and you get a "move" uevent, but it's not easy to
* connect the event to the old and new device. Device nodes are not renamed at
* all, there isn't even support for that in the kernel now.
*
* In the meantime, during renaming, your target name might be taken by another
* driver, creating conflicts. Or the old name is taken directly after you
* renamed it -- then you get events for the same DEVPATH, before you even see
* the "move" event. It's just a mess, and nothing new should ever rely on
* kernel device renaming. Besides that, it's not even implemented now for
* other things than (driver-core wise very simple) network devices.
*
* We are currently about to change network renaming in udev to completely
* disallow renaming of devices in the same namespace as the kernel uses,
* because we can't solve the problems properly, that arise with swapping names
* of multiple interfaces without races. Means, renaming of eth[0-9]* will only
* be allowed to some other name than eth[0-9]*, for the aforementioned
* reasons.
*
* Make up a "real" name in the driver before you register anything, or add
* some other attributes for userspace to find the device, or use udev to add
* symlinks -- but never rename kernel devices later, it's a complete mess. We
* don't even want to get into that and try to implement the missing pieces in
* the core. We really have other pieces to fix in the driver core mess. :)
*/
int device_rename(struct device *dev, const char *new_name)
{
struct kobject *kobj = &dev->kobj;
char *old_device_name = NULL;
int error;
dev = get_device(dev);
if (!dev)
return -EINVAL;
dev_dbg(dev, "renaming to %s\n", new_name);
old_device_name = kstrdup(dev_name(dev), GFP_KERNEL);
if (!old_device_name) {
error = -ENOMEM;
goto out;
}
if (dev->class) {
error = sysfs_rename_link_ns(&dev->class->p->subsys.kobj,
kobj, old_device_name,
new_name, kobject_namespace(kobj));
if (error)
goto out;
}
error = kobject_rename(kobj, new_name);
if (error)
goto out;
out:
put_device(dev);
kfree(old_device_name);
return error;
}
EXPORT_SYMBOL_GPL(device_rename);
static int device_move_class_links(struct device *dev,
struct device *old_parent,
struct device *new_parent)
{
int error = 0;
if (old_parent)
sysfs_remove_link(&dev->kobj, "device");
if (new_parent)
error = sysfs_create_link(&dev->kobj, &new_parent->kobj,
"device");
return error;
}
/**
* device_move - moves a device to a new parent
* @dev: the pointer to the struct device to be moved
* @new_parent: the new parent of the device (can be NULL)
* @dpm_order: how to reorder the dpm_list
*/
int device_move(struct device *dev, struct device *new_parent,
enum dpm_order dpm_order)
{
int error;
struct device *old_parent;
struct kobject *new_parent_kobj;
dev = get_device(dev);
if (!dev)
return -EINVAL;
device_pm_lock();
new_parent = get_device(new_parent);
new_parent_kobj = get_device_parent(dev, new_parent);
if (IS_ERR(new_parent_kobj)) {
error = PTR_ERR(new_parent_kobj);
put_device(new_parent);
goto out;
}
pr_debug("device: '%s': %s: moving to '%s'\n", dev_name(dev),
__func__, new_parent ? dev_name(new_parent) : "<NULL>");
error = kobject_move(&dev->kobj, new_parent_kobj);
if (error) {
cleanup_glue_dir(dev, new_parent_kobj);
put_device(new_parent);
goto out;
}
old_parent = dev->parent;
dev->parent = new_parent;
if (old_parent)
klist_remove(&dev->p->knode_parent);
if (new_parent) {
klist_add_tail(&dev->p->knode_parent,
&new_parent->p->klist_children);
set_dev_node(dev, dev_to_node(new_parent));
}
if (dev->class) {
error = device_move_class_links(dev, old_parent, new_parent);
if (error) {
/* We ignore errors on cleanup since we're hosed anyway... */
device_move_class_links(dev, new_parent, old_parent);
if (!kobject_move(&dev->kobj, &old_parent->kobj)) {
if (new_parent)
klist_remove(&dev->p->knode_parent);
dev->parent = old_parent;
if (old_parent) {
klist_add_tail(&dev->p->knode_parent,
&old_parent->p->klist_children);
set_dev_node(dev, dev_to_node(old_parent));
}
}
cleanup_glue_dir(dev, new_parent_kobj);
put_device(new_parent);
goto out;
}
}
switch (dpm_order) {
case DPM_ORDER_NONE:
break;
case DPM_ORDER_DEV_AFTER_PARENT:
device_pm_move_after(dev, new_parent);
driver core: correct device's shutdown order Now device's shutdown sequence is performed in reverse order of their registration in devices_kset list and this sequence corresponds to the reverse device's creation order. So, devices_kset data tracks "parent<-child" device's dependencies only. Unfortunately, that's not enough and causes problems in case of implementing board's specific shutdown procedures. For example [1]: "DRA7XX_evm uses PCF8575 and one of the PCF output lines feeds to MMC/SD and this line should be driven high in order for the MMC/SD to be detected. This line is modelled as regulator and the hsmmc driver takes care of enabling and disabling it. In the case of 'reboot', during shutdown path as part of it's cleanup process the hsmmc driver disables this regulator. This makes MMC boot not functional." To handle this issue the .shutdown() callback could be implemented for PCF8575 device where corresponding GPIO pins will be configured to states, required for correct warm/cold reset. This can be achieved only when all .shutdown() callbacks have been called already for all PCF8575's consumers. But devices_kset is not filled correctly now: devices_kset: Device61 4e000000.dmm devices_kset: Device62 48070000.i2c devices_kset: Device63 48072000.i2c devices_kset: Device64 48060000.i2c devices_kset: Device65 4809c000.mmc ... devices_kset: Device102 fixedregulator-sd ... devices_kset: Device181 0-0020 // PCF8575 devices_kset: Device182 gpiochip496 devices_kset: Device183 0-0021 // PCF8575 devices_kset: Device184 gpiochip480 As can be seen from above .shutdown() callback for PCF8575 will be called before its consumers, which, in turn means, that any changes of PCF8575 GPIO's pins will be or unsafe or overwritten later by GPIO's consumers. The problem can be solved if devices_kset list will be filled not only according device creation order, but also according device's probing order to track "supplier<-consumer" dependencies also. Hence, as a fix, lets add devices_kset_move_last(), devices_kset_move_before(), devices_kset_move_after() and call them from device_move() and also add call of devices_kset_move_last() in really_probe(). After this change all entries in devices_kset will be sorted according to device's creation ("parent<-child") and probing ("supplier<-consumer") order. devices_kset after: devices_kset: Device121 48070000.i2c devices_kset: Device122 i2c-0 ... devices_kset: Device147 regulator.24 devices_kset: Device148 0-0020 devices_kset: Device149 gpiochip496 devices_kset: Device150 0-0021 devices_kset: Device151 gpiochip480 devices_kset: Device152 0-0019 ... devices_kset: Device372 fixedregulator-sd devices_kset: Device373 regulator.29 devices_kset: Device374 4809c000.mmc devices_kset: Device375 mmc0 [1] http://www.spinics.net/lists/linux-mmc/msg29825.html Cc: Sekhar Nori <nsekhar@ti.com> Signed-off-by: Grygorii Strashko <grygorii.strashko@ti.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2015-07-27 17:43:01 +00:00
devices_kset_move_after(dev, new_parent);
break;
case DPM_ORDER_PARENT_BEFORE_DEV:
device_pm_move_before(new_parent, dev);
driver core: correct device's shutdown order Now device's shutdown sequence is performed in reverse order of their registration in devices_kset list and this sequence corresponds to the reverse device's creation order. So, devices_kset data tracks "parent<-child" device's dependencies only. Unfortunately, that's not enough and causes problems in case of implementing board's specific shutdown procedures. For example [1]: "DRA7XX_evm uses PCF8575 and one of the PCF output lines feeds to MMC/SD and this line should be driven high in order for the MMC/SD to be detected. This line is modelled as regulator and the hsmmc driver takes care of enabling and disabling it. In the case of 'reboot', during shutdown path as part of it's cleanup process the hsmmc driver disables this regulator. This makes MMC boot not functional." To handle this issue the .shutdown() callback could be implemented for PCF8575 device where corresponding GPIO pins will be configured to states, required for correct warm/cold reset. This can be achieved only when all .shutdown() callbacks have been called already for all PCF8575's consumers. But devices_kset is not filled correctly now: devices_kset: Device61 4e000000.dmm devices_kset: Device62 48070000.i2c devices_kset: Device63 48072000.i2c devices_kset: Device64 48060000.i2c devices_kset: Device65 4809c000.mmc ... devices_kset: Device102 fixedregulator-sd ... devices_kset: Device181 0-0020 // PCF8575 devices_kset: Device182 gpiochip496 devices_kset: Device183 0-0021 // PCF8575 devices_kset: Device184 gpiochip480 As can be seen from above .shutdown() callback for PCF8575 will be called before its consumers, which, in turn means, that any changes of PCF8575 GPIO's pins will be or unsafe or overwritten later by GPIO's consumers. The problem can be solved if devices_kset list will be filled not only according device creation order, but also according device's probing order to track "supplier<-consumer" dependencies also. Hence, as a fix, lets add devices_kset_move_last(), devices_kset_move_before(), devices_kset_move_after() and call them from device_move() and also add call of devices_kset_move_last() in really_probe(). After this change all entries in devices_kset will be sorted according to device's creation ("parent<-child") and probing ("supplier<-consumer") order. devices_kset after: devices_kset: Device121 48070000.i2c devices_kset: Device122 i2c-0 ... devices_kset: Device147 regulator.24 devices_kset: Device148 0-0020 devices_kset: Device149 gpiochip496 devices_kset: Device150 0-0021 devices_kset: Device151 gpiochip480 devices_kset: Device152 0-0019 ... devices_kset: Device372 fixedregulator-sd devices_kset: Device373 regulator.29 devices_kset: Device374 4809c000.mmc devices_kset: Device375 mmc0 [1] http://www.spinics.net/lists/linux-mmc/msg29825.html Cc: Sekhar Nori <nsekhar@ti.com> Signed-off-by: Grygorii Strashko <grygorii.strashko@ti.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2015-07-27 17:43:01 +00:00
devices_kset_move_before(new_parent, dev);
break;
case DPM_ORDER_DEV_LAST:
device_pm_move_last(dev);
driver core: correct device's shutdown order Now device's shutdown sequence is performed in reverse order of their registration in devices_kset list and this sequence corresponds to the reverse device's creation order. So, devices_kset data tracks "parent<-child" device's dependencies only. Unfortunately, that's not enough and causes problems in case of implementing board's specific shutdown procedures. For example [1]: "DRA7XX_evm uses PCF8575 and one of the PCF output lines feeds to MMC/SD and this line should be driven high in order for the MMC/SD to be detected. This line is modelled as regulator and the hsmmc driver takes care of enabling and disabling it. In the case of 'reboot', during shutdown path as part of it's cleanup process the hsmmc driver disables this regulator. This makes MMC boot not functional." To handle this issue the .shutdown() callback could be implemented for PCF8575 device where corresponding GPIO pins will be configured to states, required for correct warm/cold reset. This can be achieved only when all .shutdown() callbacks have been called already for all PCF8575's consumers. But devices_kset is not filled correctly now: devices_kset: Device61 4e000000.dmm devices_kset: Device62 48070000.i2c devices_kset: Device63 48072000.i2c devices_kset: Device64 48060000.i2c devices_kset: Device65 4809c000.mmc ... devices_kset: Device102 fixedregulator-sd ... devices_kset: Device181 0-0020 // PCF8575 devices_kset: Device182 gpiochip496 devices_kset: Device183 0-0021 // PCF8575 devices_kset: Device184 gpiochip480 As can be seen from above .shutdown() callback for PCF8575 will be called before its consumers, which, in turn means, that any changes of PCF8575 GPIO's pins will be or unsafe or overwritten later by GPIO's consumers. The problem can be solved if devices_kset list will be filled not only according device creation order, but also according device's probing order to track "supplier<-consumer" dependencies also. Hence, as a fix, lets add devices_kset_move_last(), devices_kset_move_before(), devices_kset_move_after() and call them from device_move() and also add call of devices_kset_move_last() in really_probe(). After this change all entries in devices_kset will be sorted according to device's creation ("parent<-child") and probing ("supplier<-consumer") order. devices_kset after: devices_kset: Device121 48070000.i2c devices_kset: Device122 i2c-0 ... devices_kset: Device147 regulator.24 devices_kset: Device148 0-0020 devices_kset: Device149 gpiochip496 devices_kset: Device150 0-0021 devices_kset: Device151 gpiochip480 devices_kset: Device152 0-0019 ... devices_kset: Device372 fixedregulator-sd devices_kset: Device373 regulator.29 devices_kset: Device374 4809c000.mmc devices_kset: Device375 mmc0 [1] http://www.spinics.net/lists/linux-mmc/msg29825.html Cc: Sekhar Nori <nsekhar@ti.com> Signed-off-by: Grygorii Strashko <grygorii.strashko@ti.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2015-07-27 17:43:01 +00:00
devices_kset_move_last(dev);
break;
}
put_device(old_parent);
out:
device_pm_unlock();
put_device(dev);
return error;
}
EXPORT_SYMBOL_GPL(device_move);
static int device_attrs_change_owner(struct device *dev, kuid_t kuid,
kgid_t kgid)
{
struct kobject *kobj = &dev->kobj;
struct class *class = dev->class;
const struct device_type *type = dev->type;
int error;
if (class) {
/*
* Change the device groups of the device class for @dev to
* @kuid/@kgid.
*/
error = sysfs_groups_change_owner(kobj, class->dev_groups, kuid,
kgid);
if (error)
return error;
}
if (type) {
/*
* Change the device groups of the device type for @dev to
* @kuid/@kgid.
*/
error = sysfs_groups_change_owner(kobj, type->groups, kuid,
kgid);
if (error)
return error;
}
/* Change the device groups of @dev to @kuid/@kgid. */
error = sysfs_groups_change_owner(kobj, dev->groups, kuid, kgid);
if (error)
return error;
if (device_supports_offline(dev) && !dev->offline_disabled) {
/* Change online device attributes of @dev to @kuid/@kgid. */
error = sysfs_file_change_owner(kobj, dev_attr_online.attr.name,
kuid, kgid);
if (error)
return error;
}
return 0;
}
/**
* device_change_owner - change the owner of an existing device.
* @dev: device.
* @kuid: new owner's kuid
* @kgid: new owner's kgid
*
* This changes the owner of @dev and its corresponding sysfs entries to
* @kuid/@kgid. This function closely mirrors how @dev was added via driver
* core.
*
* Returns 0 on success or error code on failure.
*/
int device_change_owner(struct device *dev, kuid_t kuid, kgid_t kgid)
{
int error;
struct kobject *kobj = &dev->kobj;
dev = get_device(dev);
if (!dev)
return -EINVAL;
/*
* Change the kobject and the default attributes and groups of the
* ktype associated with it to @kuid/@kgid.
*/
error = sysfs_change_owner(kobj, kuid, kgid);
if (error)
goto out;
/*
* Change the uevent file for @dev to the new owner. The uevent file
* was created in a separate step when @dev got added and we mirror
* that step here.
*/
error = sysfs_file_change_owner(kobj, dev_attr_uevent.attr.name, kuid,
kgid);
if (error)
goto out;
/*
* Change the device groups, the device groups associated with the
* device class, and the groups associated with the device type of @dev
* to @kuid/@kgid.
*/
error = device_attrs_change_owner(dev, kuid, kgid);
if (error)
goto out;
error = dpm_sysfs_change_owner(dev, kuid, kgid);
if (error)
goto out;
#ifdef CONFIG_BLOCK
if (sysfs_deprecated && dev->class == &block_class)
goto out;
#endif
/*
* Change the owner of the symlink located in the class directory of
* the device class associated with @dev which points to the actual
* directory entry for @dev to @kuid/@kgid. This ensures that the
* symlink shows the same permissions as its target.
*/
error = sysfs_link_change_owner(&dev->class->p->subsys.kobj, &dev->kobj,
dev_name(dev), kuid, kgid);
if (error)
goto out;
out:
put_device(dev);
return error;
}
EXPORT_SYMBOL_GPL(device_change_owner);
/**
* device_shutdown - call ->shutdown() on each device to shutdown.
*/
void device_shutdown(void)
{
struct device *dev, *parent;
wait_for_device_probe();
device_block_probing();
cpufreq_suspend();
spin_lock(&devices_kset->list_lock);
/*
* Walk the devices list backward, shutting down each in turn.
* Beware that device unplug events may also start pulling
* devices offline, even as the system is shutting down.
*/
while (!list_empty(&devices_kset->list)) {
dev = list_entry(devices_kset->list.prev, struct device,
kobj.entry);
/*
* hold reference count of device's parent to
* prevent it from being freed because parent's
* lock is to be held
*/
parent = get_device(dev->parent);
get_device(dev);
/*
* Make sure the device is off the kset list, in the
* event that dev->*->shutdown() doesn't remove it.
*/
list_del_init(&dev->kobj.entry);
spin_unlock(&devices_kset->list_lock);
/* hold lock to avoid race with probe/release */
if (parent)
device_lock(parent);
device_lock(dev);
/* Don't allow any more runtime suspends */
pm_runtime_get_noresume(dev);
pm_runtime_barrier(dev);
if (dev->class && dev->class->shutdown_pre) {
if (initcall_debug)
dev_info(dev, "shutdown_pre\n");
dev->class->shutdown_pre(dev);
}
if (dev->bus && dev->bus->shutdown) {
if (initcall_debug)
dev_info(dev, "shutdown\n");
dev->bus->shutdown(dev);
} else if (dev->driver && dev->driver->shutdown) {
if (initcall_debug)
dev_info(dev, "shutdown\n");
dev->driver->shutdown(dev);
}
device_unlock(dev);
if (parent)
device_unlock(parent);
put_device(dev);
put_device(parent);
spin_lock(&devices_kset->list_lock);
}
spin_unlock(&devices_kset->list_lock);
}
/*
* Device logging functions
*/
#ifdef CONFIG_PRINTK
static void
set_dev_info(const struct device *dev, struct dev_printk_info *dev_info)
{
const char *subsys;
memset(dev_info, 0, sizeof(*dev_info));
if (dev->class)
subsys = dev->class->name;
else if (dev->bus)
subsys = dev->bus->name;
else
return;
strscpy(dev_info->subsystem, subsys, sizeof(dev_info->subsystem));
/*
* Add device identifier DEVICE=:
* b12:8 block dev_t
* c127:3 char dev_t
* n8 netdev ifindex
* +sound:card0 subsystem:devname
*/
if (MAJOR(dev->devt)) {
char c;
if (strcmp(subsys, "block") == 0)
c = 'b';
else
c = 'c';
snprintf(dev_info->device, sizeof(dev_info->device),
"%c%u:%u", c, MAJOR(dev->devt), MINOR(dev->devt));
} else if (strcmp(subsys, "net") == 0) {
struct net_device *net = to_net_dev(dev);
snprintf(dev_info->device, sizeof(dev_info->device),
"n%u", net->ifindex);
} else {
snprintf(dev_info->device, sizeof(dev_info->device),
"+%s:%s", subsys, dev_name(dev));
}
}
int dev_vprintk_emit(int level, const struct device *dev,
const char *fmt, va_list args)
{
struct dev_printk_info dev_info;
set_dev_info(dev, &dev_info);
return vprintk_emit(0, level, &dev_info, fmt, args);
}
EXPORT_SYMBOL(dev_vprintk_emit);
int dev_printk_emit(int level, const struct device *dev, const char *fmt, ...)
{
va_list args;
int r;
va_start(args, fmt);
r = dev_vprintk_emit(level, dev, fmt, args);
va_end(args);
return r;
}
EXPORT_SYMBOL(dev_printk_emit);
static void __dev_printk(const char *level, const struct device *dev,
struct va_format *vaf)
{
if (dev)
dev_printk_emit(level[1] - '0', dev, "%s %s: %pV",
dev_driver_string(dev), dev_name(dev), vaf);
else
printk("%s(NULL device *): %pV", level, vaf);
}
printk: index: Add indexing support to dev_printk While for most kinds of issues we have counters, tracepoints, or metrics with a stable interface which can reliably be used to indicate issues, in order to react to production issues quickly we sometimes need to work with the interface which most kernel developers naturally use when developing: printk, and printk-esques like dev_printk. dev_printk is by far the most likely custom subsystem printk to benefit from the printk indexing infrastructure, since niche device issues brought about by production changes, firmware upgrades, and the like are one of the most common things that we need printk infrastructure's assistance to monitor. Often these errors were never expected to practically manifest in reality, and exhibit in code without extensive (or any) metrics present. As such, there are typically very few options for issue detection available to those with large fleets at the time the incident happens, and we thus benefit strongly from monitoring netconsole in these instances. As such, add the infrastructure for dev_printk to be indexed in the printk index. Even on a minimal kernel config, the coverage of the base kernel's printk index is significantly improved: Before: [root@ktst ~]# wc -l /sys/kernel/debug/printk/index/vmlinux 4497 /sys/kernel/debug/printk/index/vmlinux After: [root@ktst ~]# wc -l /sys/kernel/debug/printk/index/vmlinux 5573 /sys/kernel/debug/printk/index/vmlinux In terms of implementation, in order to trivially disambiguate them, dev_printk is now a macro which wraps _dev_printk. Signed-off-by: Chris Down <chris@chrisdown.name> Cc: Petr Mladek <pmladek@suse.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk> Reviewed-by: Petr Mladek <pmladek@suse.com> Tested-by: Petr Mladek <pmladek@suse.com> Acked-by: Andy Shevchenko <andy.shevchenko@gmail.com> Signed-off-by: Petr Mladek <pmladek@suse.com> Link: https://lore.kernel.org/r/959c7aed1017cb2c9de922e0a820d397e29c6a5a.1623775748.git.chris@chrisdown.name
2021-06-15 16:52:56 +00:00
void _dev_printk(const char *level, const struct device *dev,
const char *fmt, ...)
{
struct va_format vaf;
va_list args;
va_start(args, fmt);
vaf.fmt = fmt;
vaf.va = &args;
__dev_printk(level, dev, &vaf);
va_end(args);
}
printk: index: Add indexing support to dev_printk While for most kinds of issues we have counters, tracepoints, or metrics with a stable interface which can reliably be used to indicate issues, in order to react to production issues quickly we sometimes need to work with the interface which most kernel developers naturally use when developing: printk, and printk-esques like dev_printk. dev_printk is by far the most likely custom subsystem printk to benefit from the printk indexing infrastructure, since niche device issues brought about by production changes, firmware upgrades, and the like are one of the most common things that we need printk infrastructure's assistance to monitor. Often these errors were never expected to practically manifest in reality, and exhibit in code without extensive (or any) metrics present. As such, there are typically very few options for issue detection available to those with large fleets at the time the incident happens, and we thus benefit strongly from monitoring netconsole in these instances. As such, add the infrastructure for dev_printk to be indexed in the printk index. Even on a minimal kernel config, the coverage of the base kernel's printk index is significantly improved: Before: [root@ktst ~]# wc -l /sys/kernel/debug/printk/index/vmlinux 4497 /sys/kernel/debug/printk/index/vmlinux After: [root@ktst ~]# wc -l /sys/kernel/debug/printk/index/vmlinux 5573 /sys/kernel/debug/printk/index/vmlinux In terms of implementation, in order to trivially disambiguate them, dev_printk is now a macro which wraps _dev_printk. Signed-off-by: Chris Down <chris@chrisdown.name> Cc: Petr Mladek <pmladek@suse.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk> Reviewed-by: Petr Mladek <pmladek@suse.com> Tested-by: Petr Mladek <pmladek@suse.com> Acked-by: Andy Shevchenko <andy.shevchenko@gmail.com> Signed-off-by: Petr Mladek <pmladek@suse.com> Link: https://lore.kernel.org/r/959c7aed1017cb2c9de922e0a820d397e29c6a5a.1623775748.git.chris@chrisdown.name
2021-06-15 16:52:56 +00:00
EXPORT_SYMBOL(_dev_printk);
#define define_dev_printk_level(func, kern_level) \
void func(const struct device *dev, const char *fmt, ...) \
{ \
struct va_format vaf; \
va_list args; \
\
va_start(args, fmt); \
\
vaf.fmt = fmt; \
vaf.va = &args; \
\
__dev_printk(kern_level, dev, &vaf); \
\
va_end(args); \
} \
EXPORT_SYMBOL(func);
define_dev_printk_level(_dev_emerg, KERN_EMERG);
define_dev_printk_level(_dev_alert, KERN_ALERT);
define_dev_printk_level(_dev_crit, KERN_CRIT);
define_dev_printk_level(_dev_err, KERN_ERR);
define_dev_printk_level(_dev_warn, KERN_WARNING);
define_dev_printk_level(_dev_notice, KERN_NOTICE);
define_dev_printk_level(_dev_info, KERN_INFO);
#endif
2015-04-03 21:23:37 +00:00
/**
* dev_err_probe - probe error check and log helper
* @dev: the pointer to the struct device
* @err: error value to test
* @fmt: printf-style format string
* @...: arguments as specified in the format string
*
* This helper implements common pattern present in probe functions for error
* checking: print debug or error message depending if the error value is
* -EPROBE_DEFER and propagate error upwards.
* In case of -EPROBE_DEFER it sets also defer probe reason, which can be
* checked later by reading devices_deferred debugfs attribute.
* It replaces code sequence::
*
* if (err != -EPROBE_DEFER)
* dev_err(dev, ...);
* else
* dev_dbg(dev, ...);
* return err;
*
* with::
*
* return dev_err_probe(dev, err, ...);
*
* Note that it is deemed acceptable to use this function for error
* prints during probe even if the @err is known to never be -EPROBE_DEFER.
* The benefit compared to a normal dev_err() is the standardized format
* of the error code and the fact that the error code is returned.
*
* Returns @err.
*
*/
int dev_err_probe(const struct device *dev, int err, const char *fmt, ...)
{
struct va_format vaf;
va_list args;
va_start(args, fmt);
vaf.fmt = fmt;
vaf.va = &args;
if (err != -EPROBE_DEFER) {
dev_err(dev, "error %pe: %pV", ERR_PTR(err), &vaf);
} else {
device_set_deferred_probe_reason(dev, &vaf);
dev_dbg(dev, "error %pe: %pV", ERR_PTR(err), &vaf);
}
va_end(args);
return err;
}
EXPORT_SYMBOL_GPL(dev_err_probe);
2015-04-03 21:23:37 +00:00
static inline bool fwnode_is_primary(struct fwnode_handle *fwnode)
{
return fwnode && !IS_ERR(fwnode->secondary);
}
/**
* set_primary_fwnode - Change the primary firmware node of a given device.
* @dev: Device to handle.
* @fwnode: New primary firmware node of the device.
*
* Set the device's firmware node pointer to @fwnode, but if a secondary
* firmware node of the device is present, preserve it.
*
* Valid fwnode cases are:
* - primary --> secondary --> -ENODEV
* - primary --> NULL
* - secondary --> -ENODEV
* - NULL
2015-04-03 21:23:37 +00:00
*/
void set_primary_fwnode(struct device *dev, struct fwnode_handle *fwnode)
{
struct device *parent = dev->parent;
struct fwnode_handle *fn = dev->fwnode;
2015-04-03 21:23:37 +00:00
if (fwnode) {
2015-04-03 21:23:37 +00:00
if (fwnode_is_primary(fn))
fn = fn->secondary;
if (fn) {
WARN_ON(fwnode->secondary);
fwnode->secondary = fn;
}
2015-04-03 21:23:37 +00:00
dev->fwnode = fwnode;
} else {
if (fwnode_is_primary(fn)) {
dev->fwnode = fn->secondary;
/* Set fn->secondary = NULL, so fn remains the primary fwnode */
if (!(parent && fn == parent->fwnode))
fn->secondary = NULL;
} else {
dev->fwnode = NULL;
}
2015-04-03 21:23:37 +00:00
}
}
EXPORT_SYMBOL_GPL(set_primary_fwnode);
/**
* set_secondary_fwnode - Change the secondary firmware node of a given device.
* @dev: Device to handle.
* @fwnode: New secondary firmware node of the device.
*
* If a primary firmware node of the device is present, set its secondary
* pointer to @fwnode. Otherwise, set the device's firmware node pointer to
* @fwnode.
*/
void set_secondary_fwnode(struct device *dev, struct fwnode_handle *fwnode)
{
if (fwnode)
fwnode->secondary = ERR_PTR(-ENODEV);
if (fwnode_is_primary(dev->fwnode))
dev->fwnode->secondary = fwnode;
else
dev->fwnode = fwnode;
}
EXPORT_SYMBOL_GPL(set_secondary_fwnode);
driver core: add helper to reuse a device-tree node Add a helper function to be used when reusing the device-tree node of another device. It is fairly common for drivers to reuse the device-tree node of a parent (or other ancestor) device when creating class or bus devices (e.g. gpio chips, i2c adapters, iio chips, spi masters, serdev, phys, usb root hubs). But reusing a device-tree node may cause problems if the new device is later probed as for example driver core would currently attempt to reinitialise an already active associated pinmux configuration. Other potential issues include the platform-bus code unconditionally dropping the device-tree node reference in its device destructor, reinitialisation of other bus-managed resources such as clocks, and the recently added DMA-setup in driver core. Note that for most examples above this is currently not an issue as the devices are never probed, but this is a problem for the USB bus which has recently gained device-tree support. This was discovered and worked-around in a rather ad-hoc fashion by commit dc5878abf49c ("usb: core: move root hub's device node assignment after it is added to bus") by not setting the of_node pointer until after the root-hub device has been registered. Instead we can allow devices to reuse a device-tree node by setting a flag in their struct device that can be used by core, bus and driver code to avoid resources from being over-allocated. Note that the helper also grabs an extra reference to the device node, which specifically balances the unconditional put in the platform-device destructor. Signed-off-by: Johan Hovold <johan@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2017-06-06 15:59:00 +00:00
/**
* device_set_of_node_from_dev - reuse device-tree node of another device
* @dev: device whose device-tree node is being set
* @dev2: device whose device-tree node is being reused
*
* Takes another reference to the new device-tree node after first dropping
* any reference held to the old node.
*/
void device_set_of_node_from_dev(struct device *dev, const struct device *dev2)
{
of_node_put(dev->of_node);
dev->of_node = of_node_get(dev2->of_node);
dev->of_node_reused = true;
}
EXPORT_SYMBOL_GPL(device_set_of_node_from_dev);
drivers: Add generic helper to match by of_node Add a helper to match device by the of_node. This will be later used to provide wrappers to the device iterators for {bus/class/driver}_find_device(). Convert other users to reuse this new helper. Cc: Alan Tull <atull@kernel.org> Cc: Andrew Lunn <andrew@lunn.ch> Cc: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@linux.ie> Cc: "David S. Miller" <davem@davemloft.net> Cc: devicetree@vger.kernel.org Cc: dri-devel@lists.freedesktop.org Cc: Florian Fainelli <f.fainelli@gmail.com> Cc: Frank Rowand <frowand.list@gmail.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Heiner Kallweit <hkallweit1@gmail.com> Cc: Jiri Slaby <jslaby@suse.com> Cc: Jonathan Hunter <jonathanh@nvidia.com> Cc: Lee Jones <lee.jones@linaro.org> Cc: Liam Girdwood <lgirdwood@gmail.com> Cc: linux-fpga@vger.kernel.org Cc: linux-i2c@vger.kernel.org Cc: linux-spi@vger.kernel.org Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Cc: Mark Brown <broonie@kernel.org> Cc: Mathieu Poirier <mathieu.poirier@linaro.org> Cc: Maxime Ripard <maxime.ripard@bootlin.com> Cc: Moritz Fischer <mdf@kernel.org> Cc: Peter Rosin <peda@axentia.se> Cc: Rob Herring <robh+dt@kernel.org> Cc: Srinivas Kandagatla <srinivas.kandagatla@linaro.org> Cc: Thierry Reding <thierry.reding@gmail.com> Cc: Thor Thayer <thor.thayer@linux.intel.com> Cc: Wolfram Sang <wsa@the-dreams.de> Cc: "Rafael J. Wysocki" <rafael@kernel.org> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Ulf Hansson <ulf.hansson@linaro.org> Cc: Joe Perches <joe@perches.com> Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-06-14 17:54:01 +00:00
void device_set_node(struct device *dev, struct fwnode_handle *fwnode)
{
dev->fwnode = fwnode;
dev->of_node = to_of_node(fwnode);
}
EXPORT_SYMBOL_GPL(device_set_node);
int device_match_name(struct device *dev, const void *name)
{
return sysfs_streq(dev_name(dev), name);
}
EXPORT_SYMBOL_GPL(device_match_name);
drivers: Add generic helper to match by of_node Add a helper to match device by the of_node. This will be later used to provide wrappers to the device iterators for {bus/class/driver}_find_device(). Convert other users to reuse this new helper. Cc: Alan Tull <atull@kernel.org> Cc: Andrew Lunn <andrew@lunn.ch> Cc: Daniel Vetter <daniel@ffwll.ch> Cc: David Airlie <airlied@linux.ie> Cc: "David S. Miller" <davem@davemloft.net> Cc: devicetree@vger.kernel.org Cc: dri-devel@lists.freedesktop.org Cc: Florian Fainelli <f.fainelli@gmail.com> Cc: Frank Rowand <frowand.list@gmail.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Heiner Kallweit <hkallweit1@gmail.com> Cc: Jiri Slaby <jslaby@suse.com> Cc: Jonathan Hunter <jonathanh@nvidia.com> Cc: Lee Jones <lee.jones@linaro.org> Cc: Liam Girdwood <lgirdwood@gmail.com> Cc: linux-fpga@vger.kernel.org Cc: linux-i2c@vger.kernel.org Cc: linux-spi@vger.kernel.org Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Cc: Mark Brown <broonie@kernel.org> Cc: Mathieu Poirier <mathieu.poirier@linaro.org> Cc: Maxime Ripard <maxime.ripard@bootlin.com> Cc: Moritz Fischer <mdf@kernel.org> Cc: Peter Rosin <peda@axentia.se> Cc: Rob Herring <robh+dt@kernel.org> Cc: Srinivas Kandagatla <srinivas.kandagatla@linaro.org> Cc: Thierry Reding <thierry.reding@gmail.com> Cc: Thor Thayer <thor.thayer@linux.intel.com> Cc: Wolfram Sang <wsa@the-dreams.de> Cc: "Rafael J. Wysocki" <rafael@kernel.org> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Ulf Hansson <ulf.hansson@linaro.org> Cc: Joe Perches <joe@perches.com> Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-06-14 17:54:01 +00:00
int device_match_of_node(struct device *dev, const void *np)
{
return dev->of_node == np;
}
EXPORT_SYMBOL_GPL(device_match_of_node);
int device_match_fwnode(struct device *dev, const void *fwnode)
{
return dev_fwnode(dev) == fwnode;
}
EXPORT_SYMBOL_GPL(device_match_fwnode);
int device_match_devt(struct device *dev, const void *pdevt)
{
return dev->devt == *(dev_t *)pdevt;
}
EXPORT_SYMBOL_GPL(device_match_devt);
int device_match_acpi_dev(struct device *dev, const void *adev)
{
return ACPI_COMPANION(dev) == adev;
}
EXPORT_SYMBOL(device_match_acpi_dev);
int device_match_acpi_handle(struct device *dev, const void *handle)
{
return ACPI_HANDLE(dev) == handle;
}
EXPORT_SYMBOL(device_match_acpi_handle);
int device_match_any(struct device *dev, const void *unused)
{
return 1;
}
EXPORT_SYMBOL_GPL(device_match_any);